@comchatx/cx-sponsored-ads-playbook 0.0.0-stage → 0.1.1

package/research/areas/bids-and-placements.md

@@ -0,0 +1,796 @@
+# Bids and placements
+
+Read 2026-11-09 by the research seat.
+
+Tags: [official] is Amazon's own words, cited by page name and URL and read 2026-11-09 unless another date is given. [opinion] is a named practitioner, vendor or regulator, with the person or firm given. [research] is a paper, a working paper or a public research repository. [seat] is this seat's own reading: it may appear in this file but it never becomes a position. Figures (percentages, amounts, day counts, limits, minimum and maximum bids) are recorded with page and date **for human readers only** unless the line says otherwise. A skill reads the live value at run time from the operation named in *What cx must read at run time*, because Amazon changes these figures and, as this file shows several times, its own pages disagree about some of them.
+
+URLs. Ads API documentation pages are cited by their public address `https://advertising.amazon.com/API/docs/en-us/<path>`. That address serves a client-side shell, so each was read as markdown at `https://d3a0d0y2hgofx6.cloudfront.net/en-us/<path>.md`. OpenAPI specifications were read at the `cloudfront.net/openapi/...` addresses given in *Pages read*. Amazon Ads help pages are cited as `https://advertising.amazon.com/help/<ID>`. A plain GET of that address returns a shell with the article embedded as JSON (`"nodeId":"<ID>","content":"<html>"`), and the article was read from that JSON. Library guides, product pages and "what's new" announcements on `advertising.amazon.com` serve their text in the HTML and were read directly.
+
+## What this covers
+
+This file covers the price side of Amazon's sponsored ads: what a bid is, how Amazon's auction uses it to pick a winner and set a price, and every Amazon mechanism that moves the bid a seller typed before it enters an auction. In scope:
+
+- the auction as Amazon now states it: cost per click (CPC) and vCPM pricing, the generalized second-price auction with hard and soft reserves, relevance in ranking, and the 2026 FTC case about the soft reserve;
+- what "bid", "adjusted bid", "effective bid" and "maximum effective bid" mean on Amazon's pages, and the minimum and maximum bids per marketplace;
+- Sponsored Products bidding strategies: dynamic bids down only, dynamic bids up and down, fixed bids, rule-based bidding with a ROAS guardrail, and the names these carry in the API;
+- placement bid adjustments and their ranges: top of search, rest of search, product pages, Amazon Business, the browse placements of Sponsored Products across retailers, and the off-Amazon setting;
+- audience (shopper-cohort) bid boosts, Sponsored Products video adjustments, and the order in which adjustments multiply;
+- schedule bid rules (Amazon's own dayparting surface) against third-party dayparting;
+- bid recommendations: suggested bid and range, theme-based recommendations, impact metrics, Performance Estimate, the campaign and tactical recommendation types that touch bids, and how Amazon says it computes them;
+- top-of-search impression share and the placement reports;
+- Sponsored Brands bidding (automated placement bidding, placement and audience adjustments, cost control, reserve share of voice) and Sponsored Display, now called display ads (bid optimization for reach, page visits and conversions, vCPM, cost control);
+- how soon bid-relevant data arrives (hourly data, click invalidation, conversion restatement);
+- the bid writes themselves in each API (operations, body fields, media types, batch ceilings, per-item refusals) and when a written bid is not the bid the auction uses;
+- the AI shopping surfaces (Alexa for Shopping, into which Rufus and Alexa+ were merged in May 2026; prompts; Alexa+ Conversational Ads) as places a bid serves;
+- published research on position effects in sponsored search, auction design with reserves and soft floors, and bid optimization, including Amazon Science work.
+
+Out of scope here, and where it lives instead:
+
+- daily and lifetime budgets, budget rules, out-of-budget detection and pacing: `areas/budgets-and-pacing.md`;
+- which keywords and products to target, match types and their bids at set-up: `areas/keyword-and-product-targeting.md`;
+- how campaigns and ad groups are cut, which decides where a bid is set: `areas/campaign-structure.md`;
+- search-term harvesting and negatives: `areas/search-terms-and-negatives.md`;
+- launch-period bidding: `areas/launch.md`;
+- event periods, Prime Day and the event themes in bid recommendations as a calendar matter: `areas/seasons-and-deals.md`;
+- what advertising does to organic rank and total sales: `areas/ads-organic-and-total.md`;
+- reading a campaign that has gone wrong: `areas/diagnosis.md`;
+- which ad products a seller may use at all: `beyond/ad-products-and-eligibility.md`;
+- attribution windows and metric definitions (ACOS, ROAS, new-to-brand): `beyond/metrics-and-attribution.md`;
+- target ACOS and break-even economics: `beyond/goals-and-economics.md`;
+- defending branded terms and reserve share of voice as a competitive tool: `beyond/brand-defense-and-competition.md`;
+- Amazon's automation (Ads Agent, MCP server, auto-managed campaigns) as a whole: `beyond/automation-and-amazon-tools.md`;
+- the API calls themselves: `method/ads-api-surface.md`; report mechanics: `method/reporting.md`; how to tell whether a bid change worked: `method/change-measurement.md`.
+
+## Amazon's own rules and guidance
+
+Each item is marked RULE (Amazon states how its system behaves or what it accepts) or GUIDANCE (Amazon advises an advertiser). Where Amazon's pages disagree, both are recorded and the disagreement is named.
+
+### A. The auction: how the winner and the price are set
+
+- RULE: "To help choose the ads that appear in those placements and to determine the price, we usually run auctions. We run billions of auctions each day with multiple auctions often running simultaneously on a single page because of the variety of placements and ad formats. You may win one auction and not others on that page because of the unique qualities of each placement, your ad, and what else is on the page." [official] *Understand Amazon Ads auctions* (help GEXGZNBBWPZZK7P8), https://advertising.amazon.com/help/GEXGZNBBWPZZK7P8
+- RULE: ranking places "particular value on the ad's expected relevance to the customer", judged on variables including "The content surrounding where the ad will go", "The customer's search query (if applicable)", "The likelihood a customer will engage with the advertised product or service" and "The customer's shopping interests". "As a result, ads with lower bid amounts, but higher relevance to the customer, can and do win our auctions." [official] same page.
+- RULE: reserves. "Some reserves help allocate ad space by setting a bid threshold. Your bid must meet this amount to be allocated, and you'll be charged at least this amount if your ad wins. Other reserves influence your final CPC or vCPM based on factors including the likelihood that the ad will lead to a sale, predicted return on your ad spend, the value of any competing bids, placement, context, and predicted performance (e.g., the same ad placement may have a different value during a peak shopping period than a non-peak period). As a result, the price charged to you may exceed the runner-up bid but will never exceed the maximum bid you authorize." [official] same page.
+- RULE: "We will never charge you more than the maximum bid you authorize and you should only bid what you are willing to pay. Accordingly, as you enter your bids, make sure to account for applicable bid adjustments and dynamic bidding features that may increase or decrease your entered bid amount for some ad opportunities." [official] same page.
+- GUIDANCE: "Rather than trying to optimize for specific auctions, focus on setting bids that align with your overall goals and adjust them based on the balance between performance and cost that works for you." [official] same page.
+- RULE: the auction form, in Amazon's corporate statement. "Since we first introduced ads in our Store in 2006, we have priced clicks using a form of generalized “second-price” auction ... advertisers may pay less than their bid." The statement names "a “hard reserve,” which is the minimum a bid must surpass to enter an auction. The hard reserve helps cover our costs whereas soft reserves represent what we estimate to be the true market value of the ad placement." And: "Advertisers bid a maximum price for a placement. When the winning advertiser’s bid exceeds both the hard and soft reserve, they pay the soft reserve, which is less than they were willing to pay. When the winning advertiser’s bid exceeds the hard reserve but doesn't meet the soft reserve, we still grant the placement to that advertiser and they pay their bid. In no scenario does an advertiser pay more than their bid." [official] *Amazon's response to the FTC's lawsuit regarding Sponsored Ads*, aboutamazon.com, https://www.aboutamazon.com/md/company-news/amazon-ftc-sponsored-ads-lawsuit-response (undated on the page; it answers the complaint filed 2026-08-31).
+  - [seat] The statement describes the soft-reserve case only for a single winner; it does not say how the price is set when the runner-up's (relevance-weighted) bid is above the soft reserve. The help page's "may exceed the runner-up bid" means the price can sit above the second price. The two pages together describe a second-price auction in which the soft reserve can lift the price up to, but never above, the winner's own maximum.
+- RULE: candidate selection and ranking. Amazon says it reduces the eligible ads "down to less than a thousand candidates by removing less relevant ads. We then apply a ranking score to each eligible ad based on relevance and bid. Over time, our ranking formula has increasingly given greater weight to relevancy over bid amount." Relevance models "consider a variety of factors including the content surrounding the ad placement, the shopper’s search query and the likelihood a shopper will engage". "reserves are determined in real time and aren't predictable in advance by anyone, including Amazon or the advertiser." [official] same statement.
+- RULE (Amazon's figures, for human readers only): "in 2024, approximately 92% of selected Sponsored Products ads were not the highest bid"; "The mean winning advertiser’s bid is typically about the 128th bid by amount." The same statement claims that the average winning bid fell about 50% (from 2019 to 2024 in the body; "from 2019 to 2025" in its key takeaways), that CPC was flat after inflation from 2019 to 2024, that conversion rates rose 24% from 2021 to 2025, and that "eighty percent of bid changes on clicked Sponsored Products search ads occurred within one day of a prior change". It says help content was updated "to explicitly explain the use of reserve prices", and that the console has said since 2018 that a bid is a maximum. [official] same statement. These are Amazon's claims in litigation; none is independently verified here.
+- RULE: the same auction description, with retailers' reserves, applies to Sponsored Products across retailers (ads served on other retailers' sites through Amazon Retail Ad Service). Ads are ranked on "Your bid (the maximum cost per click you are willing to pay)", "How well the advertised product matches the shopper's context" and "The likelihood a customer will engage". "Retailers may set reserve pricing"; "the price charged to you may exceed the runner-up bid but will never exceed the adjusted maximum bid you authorize". "When a retailer integrates multiple demand sources, we may adjust our submitted price to compete effectively in the retailer's allocation while always respecting the adjusted maximum bid you authorize." [official] *Understand auctions for Sponsored Products ads across retailers* (help GW3CP75ZUTET4QEF), https://advertising.amazon.com/help/GW3CP75ZUTET4QEF
+- RULE: "The final CPC is usually determined by an auction and is based on your adjusted bid plus additional factors ... the auction considers factors beyond bid to determine both ad placement and final CPC." "The maximum cost per click is the highest amount an advertiser could be charged for a click, which includes their base bid plus any adjustments from settings such as dynamic bidding or cost controls." [official] *What is cost per click (CPC)?*, Amazon Ads library guide, https://advertising.amazon.com/library/guides/cost-per-click
+- RULE: "Your final cost-per-click will be based on your adjusted bid plus additional factors, but will never exceed your maximum adjusted bid." [official] *Sponsored Products best practices*, library guide, https://advertising.amazon.com/library/guides/sponsored-products-best-practices
+- GUIDANCE, simplified: the author's guide tells KDP authors that "the amount you bid, along with the relevance of your ad to the reader's query, determines whether your ad will be shown", and gives a bid-only example ($0.75 against $0.60, "you will win the auction"). [official] *An author's guide to Sponsored Products*, library guide, https://advertising.amazon.com/library/guides/authors-guide-to-sponsored-products
+  - [seat] The example leaves out relevance and reserves; the auction page and the corporate statement above supersede it.
+- RULE (the regulator's allegation, not a finding): the FTC and 22 states sued on 2026-08-31 in the Western District of Washington, alleging that since 2019 Amazon has applied an undisclosed "soft reserve" so that Sponsored Products advertisers paid their own bid "close to 80% of the time": 30–40% of the time in 2021, 70% in 2022 and about 80% in 2024 (figures for human readers only). The complaint covers Sponsored Products, Sponsored Brands and display ads; the year percentages are for Sponsored Products only. [opinion: the FTC, a regulator's untested allegation] *FTC, States Sue Amazon Over Secret Ad Surcharge Scheme*, FTC press release, 2026-08-31, https://www.ftc.gov/news-events/news/press-releases/2026/08/ftc-states-sue-amazon-over-secret-ad-surcharge-scheme (a plain curl GET returned the FTC's "Page Not Found" shell on 2026-11-09; the page was read through a fetch tool the same day). Secondary coverage gives the docket as FTC v. Amazon, W.D. Wash., 2:26-cv-03097, and reports the complaint's account that soft reserves began with Sponsored Brands in 2018, Sponsored Products in 2019 and display ads by 2023 (search-result summaries of Bloomberg Government and others, 2026-11-09; not fetched). No ruling or settlement was found by search on 2026-11-09.
+  - [seat] Amazon's 92% is about who wins; the FTC's 80% is about what the winner paid. They do not contradict each other (Ad Badger says the same, below).
+
+### B. What the bid number means, and its floor and ceiling
+
+- RULE: "A bid is what you are willing to pay for a click or view before any bid adjustments or bidding strategy. Your bidding strategy and bid adjustments then determine the bid amount. This amount is your effective bid." "If you use bid adjustments or a bidding strategy, the effective bid, can be higher or lower than your bid. The highest it can reach is your maximum effective bid. The amount you pay is your CPC or vCPM, which will not exceed the maximum effective bid." [official] *Understand bidding on Sponsored ads* (help GDS2E64R9GQ35768), https://advertising.amazon.com/help/GDS2E64R9GQ35768
+- RULE: bid types by campaign. CPC bids serve Sponsored Products, Sponsored Brands with the goal "drive page visits", and Display with "page visits" or "conversions"; vCPM bids serve Sponsored Brands "grow brand impression share" and Display. [official] same page.
+- RULE: "Cost-per-click bids that are set at the keyword level take precedence over the default cost-per-click bid set at the ad group or campaign level." [official] *Set bids in a Sponsored Products campaign* (help GTMXQWASBRZTHLF2), https://advertising.amazon.com/help/GTMXQWASBRZTHLF2
+- RULE: in the API, an ad group's `defaultBid` is "used in sponsored ads as the maximum bid during the auction", and a target's `bid` is "The maximum bid for a target". [official] Amazon Ads API v1 merged OpenAPI (`AdGroupBid`, `TargetBid`), https://d1y2lf8k3vrkfu.cloudfront.net/openapi/en-us/dest/AmazonAdsAPIALLMerged_prod_3p.json
+- RULE: cost types in the unified API are `CPC`, `CPM`, `FIXED_PRICE` and `VCPM`. [official] same specification (`CostType`).
+- RULE: minimum and maximum bids per marketplace and ad product, for human readers only. Examples from the limits page: Sponsored Products US 0.02 to 1,000 USD, UK 0.02 to 1,000 GBP, JP 2 to 100,000 JPY, IN 1 to 5,000 INR; Sponsored Brands CPC US 0.10 to 49 USD and Sponsored Brands video 0.25 to 49 USD; Sponsored Brands vCPM (brand impression share) US 8 to 5,000 USD; Sponsored Display CPC US 0.02 to 1,000 USD and vCPM 1 to 1,000 USD. The page lists Egypt's Sponsored Products range as "0.15/5.50", which reads like an error. [official] *Limits, constraints, and quotas*, Ads API reference, https://advertising.amazon.com/API/docs/en-us/reference/concepts/limits
+- RULE: the help page gives a separate table of minimum vCPM bids per country (for human readers only): Sponsored Brands "Grow Brand Impressions Share" US 8 USD for image and 12 USD for video, UK 4 and 8 GBP, DE 5 and 7 EUR, JP 560 and 800 JPY; Display 1 in most currencies, with exceptions BR 2, EG 5, IN 4, JP 100, MX 5, SA 4, SG 4 and TR 1.85. "There is a minimum bid requirement for vCPM (it varies by ad type and country). Using the minimum bids may not lead to viewable impressions." [official] *Understand bidding on Sponsored ads* (help GDS2E64R9GQ35768).
+- RULE: the API rejects a bid above the budget and a bid outside the marketplace range; the Sponsored Products bidding errors include `BID_GT_BUDGET`, `BID_OUT_OF_MARKET_PLACE_RANGE` and `BID_INVALID_PLACEMENT`. [official] Sponsored Products 3.0 OpenAPI, https://d1y2lf8k3vrkfu.cloudfront.net/openapi/en-us/dest/SponsoredProducts_prod_3p.json
+- RULE: in Performance Estimate, "you can’t save bids that are lower than the daily budget you’ve set for the campaign" [sic]. [official] *Performance Estimate (beta)* (help GHFYQ2D4LQC5ABPW), https://advertising.amazon.com/help/GHFYQ2D4LQC5ABPW
+  - [seat] Read literally this says bids lower than the budget cannot be saved, which would be backwards; the API error `BID_GT_BUDGET` says the rule is the other way round (a bid may not exceed the budget). The help sentence is very probably inverted.
+
+### C. Sponsored Products bidding strategies
+
+- RULE: "In Sponsored Products campaigns, you choose between a dynamic, fixed, or rule based bidding strategy. You can also adjust your bids by audience." [official] *Bidding strategies for Sponsored Products* (help GCU2BUWJH2W3A8Z7), https://advertising.amazon.com/help/GCU2BUWJH2W3A8Z7
+- RULE, down only: "Amazon will reduce your bids for clicks that may be less likely to convert to a sale ... We may decrease your bids by up to 100% for all placements based on performance. This could happen, for example, on a less relevant search query or on a placement that doesn’t perform well". The page lists it as ideal "When you want to optimize for conversion but set the maximum bid amount", for "New advertisers" and for "Strict budgets". [official] same page.
+- RULE, up and down: "We may increase or decrease your bids by up to 100% for all placements based on performance. Since this strategy adjusts your bid up and down in proportion to the likelihood of a conversion, it may deliver more conversions for your ad spend compared to the other 2 strategies." "Example: Amazon can adjust your bid of $1.00 up to $2.00 for all placements." Listed as ideal for "Advertising a product with good historical performance data", "Campaigns targeting highly sought-after placements" and "Flexible budgets with room for higher bids". [official] same page.
+- RULE, fixed: "Amazon will use your exact bid for all opportunities and won’t adjust your bids. Compared to dynamic bidding strategies, you may get more impressions, but fewer conversions for your ad spend with this strategy." Placement adjustments still apply to fixed bids. Listed as ideal "When you have historical campaign data and a specific target ACOS" and for "Testing new products". [official] same page.
+- RULE, the up-and-down ceiling, where Amazon's pages disagree:
+  - "We will increase or decrease your bids by up to 100% for all placements" (a $1 bid can reach $2), with the warning "Ensure you are prepared to potentially spend double". [official] *Dynamic bidding for Sponsored Products*, library guide, https://advertising.amazon.com/library/guides/dynamic-bidding-sponsored-products
+  - The API model: `AUTO_FOR_SALES` "Increases or decreases your bids in real time by a maximum of 100%". [official] Sponsored Products 3.0 OpenAPI (`SponsoredProductsBiddingStrategy`).
+  - The help page above: "up to 100% for all placements".
+  - Against these, the author's guide says up and down raises bids "up to 100% for placements at the top of the first page ... up to 50% for all other placements". [official] *An author's guide to Sponsored Products*.
+  - The unified v1 model says the same of `SALES_UP_AND_DOWN`: "Increases or decreases your bids in real time by a maximum of 100%"; its `SALES_DOWN_ONLY` "Decreases your bids in real time when your ad is less likely to convert to a sale. Bids will never increase beyond your set bid." [official] Amazon Ads API v1 merged OpenAPI (`BidStrategy`).
+  - [seat] Four current Amazon sources (help, library guide, SP 3.0 model, v1 model) say 100% for all placements; the 100%/50% split is the older rule, still on the author's guide and repeated by practitioners (BidX and Perpetua below; Trellis in 2025–2026, "according to a third-party summary of Amazon's announcement"). PPC Land (Luis Rijo, 2026-10-18) states that Amazon's current guide sets 100% for all placements "while earlier versions allowed 100% at top of search and 50% elsewhere". The live behaviour is not readable from any API field.
+  - Re-read on 2026-11-09 for the gap fill. The library guide's example is worded in the present: "imagine you have a bid of $1.00. Currently, we can increase this bid by up to 100%, so that would be a maximum of $2.00 for all placements opportunities". The help page reads "We may increase or decrease your bids by up to 100% for all placements based on performance". The author's guide still reads "Bids will be increased by up to 100% for placements at the top of the first page of shopping results and by up to 50% for all other placements". The *Set bids in a Sponsored Products campaign* help page (GTMXQWASBRZTHLF2) describes up and down without any percentage. The worked example on *Understand bid adjustments in Sponsored ads* and *Adjust Sponsored Products bids* ($1, +50% top of search = $1.50, "can further increase the bid up to $3") is a top-of-search case and fits either ceiling. [official] pages as named, read 2026-11-09.
+  - [seat] How this settles. On Amazon's own pages the count is four current sources for "100% for all placements" against one, the author's guide written for KDP authors, for 100/50. No Amazon page dates a change from 100/50 to 100/100. "The older rule" is the seat's and PPC Land's reading of the word "Currently" and of the practitioner record; Amazon does not say it. A position stating a single ceiling should name the author's guide as the dissenting Amazon page, as P-bids-and-placements-8 does.
+- RULE: the API names. `LEGACY_FOR_SALES` is "Dynamic bids - down only": "Lowers your bids in real time when your ad may be less likely to convert to a sale. Campaigns created before the release of the bidding controls feature used this setting by default." `AUTO_FOR_SALES` is up and down. `MANUAL` "Uses your exact bid and any placement adjustments you set, and is not subject to dynamic bidding." `RULE_BASED` is rule-based bidding. `OTHER` appears in read responses only. A campaign created without `dynamicBidding` gets `LEGACY_FOR_SALES`. [official] Sponsored Products 3.0 OpenAPI.
+- RULE: the unified (v1) API names Sponsored Products strategies differently: its `BidStrategy` enum lists `MANUAL`, `NEW_TO_BRAND`, `PRIORITIZE_KPI_TARGET`, `RULE_BASED`, `SALES_DOWN_ONLY`, `SALES_UP_AND_DOWN`, `SPEND_BUDGET_IN_FULL` and `USE_CAMPAIGN_STRATEGY`; not every value applies to every ad product. Its KPI enum includes `ROAS` and `TOP_OF_SEARCH_IMPRESSION_SHARE`. [official] Amazon Ads API v1 merged OpenAPI.
+- RULE: the history of bid controls. On the retired `/v1/sp` API, "Bid+" (a premium top-of-search adjustment) was replaced by bidding controls; Bid+ campaigns were converted to down only with a +50% top-of-search adjustment; at that time `placementTop` and `placementProductPage` accepted 0–900%; "Bid controls apply only to Sponsored Products". [official] *Bid controls* (legacy reference), https://advertising.amazon.com/API/docs/en-us/reference/1/bid-controls
+- GUIDANCE: when to use up and down, in the library guide: the objective is sales; winning top placements; high-performing campaigns; excess inventory; deal ASINs. "review your bids often, about once every two weeks". "We recommend that you limit your changes while testing strategies, so that you can attribute the difference in performance to a specific change." [official] *Dynamic bidding for Sponsored Products*.
+- GUIDANCE: for automatic campaigns, "dynamic bidding up and down ... or, if ... ROAS, then try a dynamic down only"; for manual keyword targeting, "setting the highest bid on the exact match, a lower on the phrase match, and the lowest bid on the broad match"; "Test different match types for at least 1-2 weeks"; "you won’t be bidding against yourself by adding different match types on the same keyword". [official] *Targeting with Sponsored Products*, library guide, https://advertising.amazon.com/library/guides/targeting-with-sponsored-products
+  - [seat] "won’t be bidding against yourself" is a claim about the auction (one advertiser's own ads do not compete for the same slot); Amazon does not say which of the advertiser's eligible targets enters when several match one query. That question belongs to `areas/keyword-and-product-targeting.md`.
+- GUIDANCE: fixed bids "may get you more impressions but fewer clicks" (author's guide wording; the help page says "fewer conversions"); "Wait about two weeks after the launch of your campaign and then try getting more advanced with your bidding strategy", then "check in every two weeks after that". [official] *An author's guide to Sponsored Products*.
+- RULE: Amazon's Ad Lab video of 2026-10-15 says "With fixed bidding, your bid stays exactly where you set it" and that neither fixed nor dynamic bidding is better in every case; it published no test results. [official, through a secondary report] *Amazon Ad Lab names four bidding metrics but publishes no test results*, PPC Land (Luis Rijo), 2026-10-18, https://ppc.land/amazon-ad-lab-names-four-bidding-metrics-but-publishes-no-test-results/ , citing https://www.youtube.com/watch?v=CfoPTLaOVXo (the video itself was not watched).
+
+### D. Placement adjustments in Sponsored Products
+
+- RULE: what each placement is. "Top of search (first page): Up to the first 5 positions in the top row of the first page search results on desktop, and up to the first 4 positions on mobile." "Rest of search: the middle or bottom of shopping results and the second page of shopping results and beyond." "Product pages: Sponsored Products ads on the product details page, and all nonshopping placements such as the add-to-cart page." "Amazon business placements: all Sponsored Products placements that appear on Amazon Business including Top of Search, Rest of Search, and Amazon Business product pages." [official] *Understand bid adjustments in Sponsored ads* (help GQ4R4FR4H56JL6V8), https://advertising.amazon.com/help/GQ4R4FR4H56JL6V8 ; the same four definitions are on *Set bids in a Sponsored Products campaign* (help GTMXQWASBRZTHLF2) and *Adjust Sponsored Products bids* (help GYYZVM7LGSRYGWV5), https://advertising.amazon.com/help/GYYZVM7LGSRYGWV5
+- RULE: placement adjustments only increase. "Bid adjustments let you increase your bids when your ad is eligible to appear in specific placements, to specific audiences, or with Sponsored Product Videos. You set a percentage increase that applies on top of your base bid when certain conditions are met." "The specified increase works in addition to your selected bidding strategy (dynamic or fixed)." [official] *Understand bid adjustments in Sponsored ads*.
+- RULE: the arithmetic with dynamic bidding. "You set a dynamic bid of $1 for the keyword “camera” and set a 50% bid adjustment for top of search. When your ad is eligible to appear in the top of search against “camera” your bid will increase to $1.5. Dynamic bidding (up and down) can further increase the bid up to $3." [official] *Understand bid adjustments in Sponsored ads* and *Adjust Sponsored Products bids*.
+  - [seat] So dynamic bidding multiplies the placement-adjusted bid, not the base bid: the maximum effective bid under up and down is base × (1 + placement%) × 2.
+- RULE: range. Sponsored Products placement adjustments are percentages from 0 to 900 (`PlacementBidding.percentage`, int32, 0–900); the campaign create and update model sets no item limit on `placementBidding` (the 3-entry maximum in the model belongs only to the `SponsoredProductsNewCampaignDynamicBidding` shape used by target promotion groups, and to the placement adjustments of a new-ad-group bid recommendation request); the placements are `PLACEMENT_TOP`, `PLACEMENT_REST_OF_SEARCH`, `PLACEMENT_PRODUCT_PAGE` and `SITE_AMAZON_BUSINESS`. The model's example: "a 50% Top of Search adjustment on a $1.00 bid would increase the bid to $1.50 ... A further 100% Amazon Business adjustment would increase the bid to $3.00 for the Amazon Business Top of Search placement and to $2.00 for all other Amazon Business placements." [official] Sponsored Products 3.0 OpenAPI (`SponsoredProductsPlacement`, `SponsoredProductsPlacementBidding`).
+- RULE: rest of search became adjustable on 2023-09-21, as a beta: "up to 900%, same as the other two placements", for all targeting types and bidding strategies. [official] *Amazon Ads advanced tools release notes* (the Ads API release notes page), 2023-09-21 (API and bulksheets, beta; previewed 2023-08-06), https://advertising.amazon.com/API/docs/en-us/release-notes/index ; and *Improve your campaign performance on rest of search placements using Sponsored Products rest of search bid adjustment control*, what's new, dated 2024-01-09 (the console announcement, no "beta"), https://advertising.amazon.com/resources/whats-new/improve-campaign-performance
+- RULE: the Amazon Business adjustment arrived on 2024-10-21: "a singular adjustment ... up to 900% across all placements on Amazon Business", in the US, CA, MX, DE, UK, FR, IT, ES, IN and JP, with reporting from 2024-09-05 through the `campaignSite` filter `AmazonBusiness` on `spCampaigns` grouped by `campaignPlacement`. Amazon's claims in the note (internal 2024 data, for human readers only): business customers are "~3x more likely to purchase", have a "~50% lower return rate", and give "2-3x higher return on ad spend". [official] Ads API release notes.
+- RULE: Amazon-Business-only campaigns. `siteRestrictions` with `AMAZON_BUSINESS` restricts a campaign to Amazon Business. "If this field is not set, ads from the campaign will appear on Amazon - including both Amazon retail and Amazon Business - as well as select sites and apps off Amazon." The model says `AMAZON_BUSINESS` exists only in Amazon Business marketplaces (US, CA, MX, UK, DE, FR, IT, ES, IN, JP, AU) and `AMAZON_HAUL` only in the US; "siteRestrictions cannot be changed post campaign creation"; "siteRestrictions don’t support shopperCohortBidding setting, SITE_AMAZON_BUSINESS placementBidding setting and offAmazonSettings"; "Only AMAZON_BUSINESS option is ready for use at the moment." [official] Sponsored Products 3.0 OpenAPI (`SponsoredProductsCreateCampaign.siteRestrictions`); the release notes date the Sponsored Products option 2025-04-24 and the Sponsored Brands option 2025-06-12.
+- RULE: in Sponsored Products across retailers, "you cant adjust bids for Amazon Business placements, but have two additional placement options": "Top of browse" and "Rest of browse". [official] *Adjust Sponsored Products bids* (help GYYZVM7LGSRYGWV5). The v1 `Placement` enum carries `TOP_OF_BROWSE` and `REST_OF_BROWSE` only for the `OTHER_RETAILERS` retailer scope, beside `TOP_OF_SEARCH`, `REST_OF_SEARCH`, `PRODUCT_PAGE`, `SITE_AMAZON_BUSINESS` and `HOME_PAGE`. [official] Amazon Ads API v1 merged OpenAPI.
+- RULE: the v1 descriptions: `TOP_OF_SEARCH` is "top row of the first-page search results"; `REST_OF_SEARCH` is the "middle or the bottom of the first-page search results. Also refers to ads on the second page of search results and beyond"; `PRODUCT_PAGE` is the "product detail page, and all nonsearch placements such as the add-to-cart page". `placementBidAdjustments` holds at most 5 entries and is "Not supported for Sponsored Brands campaigns using the SALES_UP_AND_DOWN bid strategy". [official] same specification.
+- GUIDANCE: "Adjusting bids by placement is best for advertisers who have observed that certain placements have led to conversions, sales, or other desired activities in the past. You can view placement metrics at the campaign level on the bid adjustment tab in campaign manager." "We recommend this strategy for experienced advertisers who have insights about where they want their ads to serve." [official] *Understand bid adjustments in Sponsored ads*; *Set bids in a Sponsored Products campaign*.
+- GUIDANCE: "To help increase the top-of-search impression share for your Sponsored Products campaigns, adjust bids by placement". [official] *Top-of-search impression share metric*, what's new, 2021-09-22, https://advertising.amazon.com/resources/whats-new/top-of-search-impression-share-metric
+
+### E. Audience bid boosts, video adjustments, and the order in which adjustments multiply
+
+- RULE: Sponsored Products audience boosts. "Select up to 10 audiences per campaign, each with an independent bid boost percentage. When the shopper is a member of one of your selected audience(s), your bid increases by the percentage you set for that audience(s)." The two audience kinds are "Amazon audiences: Prebuilt audience segments informed by first-party shopping and streaming signals (in-market, lifestyle, interests, and life events)" and "Amazon Marketing Cloud (AMC) custom audiences". "Advertisers outside of the US marketplace can only select up to 1 audience at this time." [official] *Adjust Sponsored Products bids* (help GYYZVM7LGSRYGWV5); *Bidding strategies for Sponsored Products* (help GCU2BUWJH2W3A8Z7).
+- RULE: several audiences on one shopper. "If a shopper qualifies for more than 1 of your boosted audiences, only the highest bid boost applies. Boosts do not stack across audiences. In the case of a tie, the impression is attributed to either audience at random. Attribution is assigned to the winning audience at impression time and is not split across audiences." [official] *Adjust Sponsored Products bids*.
+- RULE: different kinds stack multiplicatively. "Bid adjustments of different types can stack. If you set a placement boost and an audience boost, and a single impression qualifies for both, both adjustments apply together to determine your final bid. The winning audience boost combines with any applicable placement or video boost." The example: $1.00 + 50% top of search = $1.50; $1.50 + 60% audience = $2.40; the 30% audience does not apply. [official] *Adjust Sponsored Products bids*; *Understand bid adjustments in Sponsored ads* (help GQ4R4FR4H56JL6V8).
+- RULE: Sponsored Products video adjustment. "Enter a bid adjustment between 0 and 900% in the "bid adjustment for video ads" box"; it can be set only after the campaign is saved, and "Increases apply on top of placement adjustments" [official] *Set bids in a Sponsored Products campaign* (help GTMXQWASBRZTHLF2). "Sponsored Products video is only available in the US and is currently not available for Books." The worked chain: $1.00, +50% rest of search = $1.50, +50% custom audience = $2.25, +100% video = $4.50. [official] *Adjust Sponsored Products bids*; *Understand bid adjustments in Sponsored ads*. In the v1 API this is `creativeBidAdjustments` (at most 2 entries, "Range of bid adjustment value would be 0:900") with the creative type `SPOTLIGHT` ("SPOTLIGHT Video Asset"). [official] Amazon Ads API v1 merged OpenAPI.
+- RULE: in the Sponsored Products 3.0 API the audience boost is `shopperCohortBidding`: at most 10 entries, type `AUDIENCE_SEGMENT` only, exactly one audience segment per entry, percentage 0 to 900, a "percentage of the original bid including any other bid adjustments such as placementBidding". The segment type field (`BEHAVIOR_DYNAMIC`, `SPONSORED_ADS_AMC`) is deprecated and inferred from the audience id, which comes from the targetable-entities listing. [official] Sponsored Products 3.0 OpenAPI (`SponsoredProductsShopperCohortBidding`).
+- RULE: the v1 `BidSettings` object holds `bidStrategy` and `bidAdjustments`; the `BidAdjustments` object carries `placementBidAdjustments` (at most 5), `audienceBidAdjustments` (at most 10), `shopperSegmentBidAdjustments` ("Legacy SB field (marked for deprecation)", at most 2, segment enum `NEW_TO_BRAND`) and `creativeBidAdjustments` (at most 2). [official] Amazon Ads API v1 merged OpenAPI (`BidSettings`, `BidAdjustments`, `ShopperSegment`).
+- RULE: dates. AMC audiences as a bid control for Sponsored Products and Sponsored Brands: 2024-10-14 (with the AMC column `matched_behavior_segment_ids`); audience bid boosting reports: 2025-01-24; Amazon-built audiences for Sponsored Products in preview, 2025-03-12 ("Purchased brand products", "Clicked or added brand product to cart", "Affinity for similar products"); a Sponsored Brands "New-to-brand shoppers" audience (not purchased from the brand in 12 months), 2025-03-24, with two more Sponsored Brands audiences on 2025-08-22. [official] Ads API release notes.
+- RULE: reporting on boosts. The usage example requests the v3 audience report types `spAudiences` and `sbAudiences` grouped by `campaign_bid_boost_segment`, with `segmentName` and `segmentClassCode` among the columns; its illustrative case is a 25% boost. [official] *Sponsored Products & Brands: Audience Bid Boosting*, Ads API knowledge hub usage example (tagged Youssry Abubakr; January 2026 per its path), https://advertising.amazon.com/API/docs/en-us/knowledge-hub/blogs/usage-examples/2026-01-sp-sb-bid-boosting
+- RULE: an Amazon-Business-only site selection removes two adjustments: "With Amazon Business Sites selection, you won't have access to audience cohort bid adjustments or Amazon Business placement bid adjustments." The site selection cannot be changed once the campaign is launched. [official] *Sites selection for Sponsored Ads campaigns* (help GJGYD2FKNF3MUWBQ), https://advertising.amazon.com/help/GJGYD2FKNF3MUWBQ
+  - [seat] The help page lists Amazon Business sites in US, CA, UK, FR, DE, IT, ES, IN, JP and MX; the API model adds AU. Read the live eligibility rather than either list.
+- [seat] The order, assembled from the pages above: base bid (keyword or target bid, else ad group default) × (1 + placement%) × (1 + highest qualifying audience%) × (1 + video%) gives the adjusted bid; the bidding strategy then moves that adjusted bid (down only: to as low as zero; up and down: up to ×2; rule-based: up to ×5, see G); schedule bid rules add their percentage on top of placement adjustments (see H). Amazon gives each step separately and in examples; no single Amazon page writes the whole chain.
+
+### F. Placements beyond the search page: off-Amazon, creators, prompts, Alexa for Shopping (formerly Rufus) and Alexa+
+
+- RULE: "Sponsored Products off-Amazon advertising is available for advertisers in Belgium, Brazil, Canada, Egypt, France, Germany, India, Ireland, Italy, Mexico, Netherlands, Poland, Saudi Arabia, Spain, Sweden, Türkiye, United Arab Emirates, United Kingdom, and the United States." "Your ads run on the targeting, bid, and budget you already set — no extra setup required." [official] *Understand Sponsored Products off-Amazon advertising* (help GYTD2Z3SYMAAMVXA), https://advertising.amazon.com/help/GYTD2Z3SYMAAMVXA
+- RULE: three routes off Amazon. "Third-party sites and apps: On third-party sites and apps, your ad is matched to a placement based on relevance and your adjusted bid. If selected, Amazon submits your ad to the publisher, where it competes with ads from other sources. The publisher decides which ads appear and in what positions." "Third-party sites and apps with Amazon investment: For some placements, Amazon may also select your product based on factors like price competitiveness, category trends, and product reviews, and invests its own marketing funds alongside your campaign". "Creator content: Creators in the Amazon Influencer and Associates Program ... can select your campaign to feature in their product reviews, editorial content, and buying guides." [official] same page.
+- RULE: off-Amazon bids. "Your maximum bid applies to every click, whether on or off Amazon, and every click counts toward the same campaign budget. Your cost-per-click (CPC) never exceeds your maximum bid, and your total spend never exceeds your budget." "Bid adjustments for Top of Search, Rest of Search, and Product Pages do not apply to off-Amazon placements. Dynamic bidding and rule-based bidding still apply." [official] same page.
+- RULE: how the off-Amazon CPC is set. Third-party sites and apps: "Amazon adjusts your bid using predicted click-through and conversion rates, and submits your ad to the publisher. The resulting CPC covers what Amazon pays the publisher (supply cost) plus Amazon’s revenue for delivering the placement." With Amazon investment: "Amazon calculates your CPC using factors like maximum bid and expected performance." Creator content: "Amazon calculates your CPC based on your maximum bid, product's expected performance and sales the ad generates. Creators may receive a portion of the CPC from qualifying clicks. You're only charged when a shopper clicks on a creator's recommendation and is taken to a detail page or adds to cart." [official] same page.
+- RULE: the off-Amazon switch. In the console: "Increase reach off-Amazon (default)" or "Limit reach to Amazon: Ads appear only on Amazon owned-and-operated properties, including creator storefronts." In the API, `offAmazonSettings.offAmazonBudgetControlStrategy` is `MAXIMIZE_REACH` ("Allow ad placements both on and off Amazon") or `MINIMIZE_SPEND` ("Restrict ad placements to Amazon owned-and-operated properties only"). A bulksheet column "Off-Amazon ad serving" was added in the US on 2026-06-08. [official] same help page; Sponsored Products 3.0 OpenAPI; Ads API release notes.
+- RULE: off-Amazon reporting. Performance appears in the Placement report, in the API reports "with the Placement Classification metric", and in console reports. "When shoppers engage with creator content off Amazon that leads to a product detail page, this activity appears as "Off Amazon" in the Placement Report. When shoppers engage with a creator's storefront on Amazon that leads to a product detail page or an add to cart, this activity appears as "Other – On Amazon."" [official] same help page.
+- RULE: the first publisher partners (2023-10-25) were Pinterest, BuzzFeed, Hearst, Raptive and Ziff Davis; "Reporting will be available in the Sponsored Products placement report ... placementClassification". [official] *Helping customers discover new selection with Sponsored Products on premium apps and websites*, what's new, https://advertising.amazon.com/resources/whats-new/sponsored-products-appear-on-publishers
+- RULE: prompts. "Prompts are an AI-powered extension to Sponsored Products and Sponsored Brands campaigns"; available in the US; "Prompts are charged on a cost-per-click (CPC) basis using your existing campaign budget"; "All existing and new Sponsored Products and Sponsored Brands campaigns in the US are automatically enrolled in prompts"; individual prompts can be paused; opting out means contacting Amazon. [official] *Understand prompts on Sponsored Products and Sponsored Brands* (help GLWNW26EWQZW46WZ), https://advertising.amazon.com/help/GLWNW26EWQZW46WZ
+  - [seat] The prompts page says nothing about which placement adjustment, if any, applies to a prompt click. Open question.
+- RULE: Rufus placements (2024-09-05, regions `US`): ads may appear in Rufus-related placements, and "Your campaign reports will not include Rufus metrics". Alexa+ Sponsored Products (2025-07-15): "only available on the Echo Show"; "Performance metrics will be consolidated within existing reporting". [official] Ads API release notes.
+  - [seat] The notes say Rufus metrics are excluded from campaign reports and Alexa+ metrics are "consolidated within existing reporting"; neither note says how those clicks are priced against placement adjustments. The prompts page, by contrast, says prompts with clicks "will appear in existing Sponsored Products and Sponsored Brands reporting".
+  - The two notes above are dated history: Amazon has since merged the two assistants (next bullets).
+- RULE: the merger. On 2026-05-13 Amazon announced "Amazon brings together Rufus and Alexa+ to create ‘Alexa for Shopping’ on the Amazon Shopping app and website", available to US customers on the app, the website and Echo Show; customers "ask questions directly in the main Amazon search bar" and get "category and product insights in search results and on product pages". [official] *Meet Alexa for Shopping, your personalized, agentic AI assistant on Amazon*, aboutamazon.com (Amazon Staff, `datePublished` 2026-05-13), https://www.aboutamazon.com/news/retail/alexa-for-shopping-ai-assistant . The Rufus page now opens "On May 13, 2026, Rufus was renamed Alexa for Shopping." [official] *'Amazon Rufus' AI experience comes to the Amazon Shopping app*, aboutamazon.com, https://www.aboutamazon.com/news/retail/amazon-rufus . Both read 2026-11-09.
+- RULE: the ad surfaces after the merger, in Amazon Ads' words (news post of 2026-06-11; page title *What agentic shopping means for advertising*, headline "Agentic shopping is here. What does that mean for advertisers?"):
+  - Sponsored ads in Alexa for Shopping: "Existing sponsored ads campaigns are automatically eligible to serve in Alexa for Shopping, helping advertisers seamlessly reach engaged audiences during relevant conversations with Alexa for Shopping, without additional campaign setup."
+  - Prompts: "Sponsored Products and Sponsored Brands prompts are AI-powered conversational ads that appear in shopping results and product pages, opening product-related conversations with Alexa for Shopping." "Existing sponsored ads campaigns are automatically eligible to be optimized as a prompt." Amazon's figures (for human readers only): "Nearly 20% of shoppers who interact with a prompt continue the conversation about that brand, and adding prompts to a Sponsored Brands ad drives a 6% increase in conversions."
+  - On Alexa+ and Alexa-enabled devices: "Alexa+ Conversational Ads were the industry's first conversational ad experience. They introduce brand messaging into personalized, natural dialogue on Alexa-enabled devices. These ads surface when relevant to a customer's conversation and intent, available in audio, static. Advertisers with sponsored ads campaigns are automatically enrolled." Prime Video Sponsored Tiles are the entertainment counterpart on eligible Echo Show devices.
+  - Measurement: "closed-loop measurement that tracks impressions all the way to conversions".
+  [official] *What agentic shopping means for advertising*, Amazon Ads news, 2026-06-11, https://advertising.amazon.com/library/news/agentic-shopping-advertising , read 2026-11-09.
+- RULE: what Amazon does not say. Neither the 2026-06-11 post, nor the prompts help page (re-read 2026-11-09), nor the *Prompt Ad Extension reports* guide says which bid, which placement adjustment (top of search, rest of search, product pages), audience boost or bidding strategy applies to an impression or click in Alexa for Shopping, in a prompt, or in Alexa+ Conversational Ads. The prompts page says only "Prompts are charged on a cost-per-click (CPC) basis using your existing campaign budget". The release notes index read 2026-11-09 has no entry for Alexa for Shopping; its last entries on the assistants are the Rufus note of 2024-09-05 and the Alexa+ note of 2025-07-15. The `spPromptAdExtension` and `sbPromptAdExtension` v3 reports (US filter only, 90-day range, 95-day retention, `SUMMARY` or `DAILY`, grouped by `promptAdExtension`) carry `promptText`, `costPerClick`, `cost` and conversions, and no placement or bid column. The v3 `placementClassification` column ("The page location where an ad appeared") and the v1 `placementClassification.value` dimension publish no list of values, so no page read says whether an Alexa for Shopping click is labelled in the placement report. [official] *Understand prompts on Sponsored Products and Sponsored Brands* (help GLWNW26EWQZW46WZ); *Prompt Ad Extension reports*, Ads API reporting v3 guide, https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/report-types/prompt-ad-extension ; *Reporting columns (version 3)*; *Placement classification (Supply)*; Ads API release notes; all read 2026-11-09.
+  - [seat] So for a bid decision: an Alexa for Shopping or Alexa+ click is paid from the same campaign and (by the prompts page) at CPC, but no Amazon page says the placement multipliers in D reach it. A placement-report row that rises after May 2026 cannot be attributed to the assistant from Amazon's documentation alone.
+
+### G. Rule-based bidding with a ROAS guardrail
+
+- RULE: "you set a rule with a performance goal (example: Return on Ad Spend or ROAS). Amazon then adjusts your base bids for each ad impression, increasing or decreasing them as needed to try to achieve your performance target. Amazon does not guarantee that we’ll hit the performance target that you set." [official] *Bidding rules for Sponsored Products* (help GDLQ5D2BNFCAU3TW), https://advertising.amazon.com/help/GDLQ5D2BNFCAU3TW
+- RULE: the ceiling. "rule-based bidding can increase your bids by up to 5 times your adjusted bid amount (if any adjustments are applied to your original bid)". The page's table: $1.00 base, top of search +10% gives $1.10 and a "Max bid caculated" of $5.50; rest of search and detail page at 0% give $1.00 and $5.00. [official] same page.
+- RULE: eligibility, in help: Sponsored Products only; the campaign must meet the marketplace's minimum daily budget; any targeting type (automatic, keyword, product); "You can specify a ROAS target in campaign settings. We recommend you should wait at least two weeks for the campaign to get enough data before specifying a ROAS target. If not specified, Amazon Ads will automatically determine an appropriate ROAS target to maximize sales." "Rule based bidding campaign eligibility criteria may change". [official] same page.
+- RULE: eligibility, in the API guide (stricter than help): the campaign has run at least 10 days and has at least 10 conversions in the last 30 days; Amazon may "adjust your base bids up and down to increase conversions"; "Amazon Ads does not guarantee the ROAS guardrail". "The guardrail is applied at the campaign level and the ROAS guardrail is based on the previous 21 days"; the guardrail recommendation "is based on your historical campaign information and a group of similar advertisers". The rule is switched off automatically if the 20-day ROAS falls below 30% of the ROAS over the 20-to-40-day period, and the campaign's previous strategy and bids return. Minimum daily budgets (for human readers only): US, UK, DE, FR, IT, ES, CA, AU, AE, NL, BE and SG 10; IN 300; JP 600; MX 50; BR 20; EG 50; SA 20; SE 50; PL 30; TR 50 (local currency). [official] *Sponsored Products Rule-based bidding overview*, Ads API guide, https://advertising.amazon.com/API/docs/en-us/guides/rules/bidding-rules/overview
+  - [seat] Help says a ROAS target is optional and Amazon sets one if absent; the API guide requires a ROAS value in the rule. These may be the console's newer behaviour and the API's older contract. Read the eligibility operation rather than either page.
+- RULE: the API surface. `POST` and `PUT /sp/rules/campaignOptimization`, `GET` and `DELETE /sp/rules/campaignOptimization/{id}`, `POST /sp/rules/campaignOptimization/state` and `POST /sp/rules/campaignOptimization/eligibility` (which, for an eligible campaign, returns ROAS and bid guardrail guidelines); recurrence `DAILY` only; action `ADOPT` only; with metric `ROAS` the comparison is `GREATER_THAN` or `EQUAL_TO`; the examples use rule type `BID`. [official] *Getting started with the Sponsored Products rule-based bidding API*, Ads API guide, https://advertising.amazon.com/API/docs/en-us/guides/rules/bidding-rules/getting-started . The content type `application/vnd.optimizationrules.v1+json`, the state operation's "Recommended refresh frequency is once a day", and `RuleConditionMetric` ("ROAS is the only supported metric", although the enum also lists `AVERAGE_BID`) are in the [official] Sponsored Products 3.0 OpenAPI.
+- RULE: rule-based bidding and schedule bid rules exclude each other. "You can’t apply schedule bid rules to campaigns that have Rule-based bidding active. Schedule bid rules will be on hold if you change your bidding strategy to Rule-based bidding." [official] *Bidding rules for Sponsored Products*.
+- GUIDANCE: rule-based bidding is listed as ideal for "Large catalogs requiring automated management", "Experienced sellers with established performance history" and "Campaigns with clear ROAS objectives". [official] *Bidding strategies for Sponsored Products* (help GCU2BUWJH2W3A8Z7).
+- RULE: in 2026 no Amazon page read here offers a "target ROAS" strategy for Sponsored Products beyond rule-based bidding; the v1 enum carries `PRIORITIZE_KPI_TARGET` ("Optimizes bidding to achieve the KPI target specified") with a KPI list that includes `ROAS`, and `SPEND_BUDGET_IN_FULL` ("Prioritize spending full budget, while maximizing performance"), neither with a Sponsored-Products-specific description. [official] Amazon Ads API v1 merged OpenAPI. A vendor reports a US beta of "Automatically optimizing ROAS within budget" (see *What practitioners add*); no Amazon page read here confirms it.
+
+### H. Schedule bid rules (Amazon's dayparting surface)
+
+- RULE: "Schedule bid rules is a feature available for Sponsored Products campaigns that allows you to schedule bid increases for specific times of the day, days of the week, date ranges or high traffic events (example: Black Friday)." They apply to "Dynamic bids - up and down", "Down only" and "Fixed". [official] *Bidding rules for Sponsored Products* (help GDLQ5D2BNFCAU3TW); *Apply Schedule bid rules for Sponsored Products* (help G7JFAU9C7PHA3Q68), https://advertising.amazon.com/help/G7JFAU9C7PHA3Q68
+- RULE: they only increase. The action operator enum holds only `INCREMENT`, the unit only `PERCENT`, and the value is "An integer between 1 & 100, representing the percent increase on base bid". [official] Sponsored Products 3.0 OpenAPI (`OptimizationRulesAPISwaggerActionDetails`, `OptimizationRulesAPISwaggerRuleActionOperator`). The guide describes schedule bid rules as rules that "automatically increase bids". [official] *Scheduled bid rules*, Ads API guide, https://advertising.amazon.com/API/docs/en-us/guides/rules/bidding-rules/scheduled-bid-rules
+- RULE: rules add, then sit on top of placement adjustments. "Schedule bid rules are add-on. Example: On campaign A you set Rule 1 for Mondays between 9am-11am with a bid increase of 20%. Then, for the same campaign, you set a daily rule for Mondays called Rule 2 with a bid increase of 10%. This means that on Mondays between 9am-11am your bids on campaign A will increase by 30%." "Schedule bid rules will be applied on top of your placement adjustment. Example: ... down only bidding strategy, top-of-search placement adjustment of 10%, bid of $1.00, and you apply a Schedule bid rule to increase your bid by 20% on Mondays. On Monday for top of search impressions, your bid will be $1.30 and $1.20 for all other placements". [official] *Bidding rules for Sponsored Products*.
+  - [seat] $1.00 × (1 + 10% + 20%) = $1.30 means the schedule percentage is added to the placement percentage, not multiplied with it. That differs from the multiplicative stacking of placement and audience adjustments in E. The API guide's own example stacks schedule rules additively too: 10% daily + 20% Black Friday = 30%, plus 20% in a four-hour window = 50%.
+- RULE: the API. Create the rule with `POST /sp/rules/optimization` (content types `application/vnd.spoptimizationrules.v1+json` and `v2`), attach it with `POST /sp/campaigns/{campaignId}/optimizationRules`, find rules with `POST /sp/rules/optimization/search`, update with `PUT`. `ruleCategory` `BID`, `ruleSubCategory` `SCHEDULE`; recurrence `DAILY` or `WEEKLY` with `daysOfWeek` and `timesOfDay` in `HH:00` (at most one window, per the model); a duration by start and end date or by `eventId` (events from `POST /sp/v1/events`). "Whenever a new schedule bid rule is created or an existing rule is updated, the changes will take effect within 24 hours. A maximum of 20 schedule bid rules can be specified for a campaign. Schedule bid rules are applied on top of 'Placement adjustment' and 'Bidding strategies'." [official] *Scheduled bid rules*, Ads API guide; the `timesOfDay` limit (`maxItems` 1) and the v2 content type are in the Sponsored Products 3.0 OpenAPI.
+- RULE: the v2 rule model also defines a `PERFORMANCE` subcategory (`OptimizationRulesAPISwaggerRuleSubCategoryV2`) with attributes `ACOS`, `CLICKS`, `CPC`, `CTR`, `CVR`, `IMPRESSIONS`, `ORDERS`, `ROAS`, `SALES` and `SPEND`, a targeting `lookbackDays` of 3 to 65 and expression types `BROAD`, `EXACT`, `EXPANDED` and `PHRASE`; its action operator is still `INCREMENT` only. [official] Sponsored Products 3.0 OpenAPI (the guide does not describe this subcategory).
+- RULE: availability when announced (2023-11-13): US, MX, CA, UK, DE, ES, FR, IT and IN; hours of the day for schedule budget rules followed on 2023-11-16. [official] Ads API release notes.
+- RULE: the hourly campaign report that goes with them. "To understand campaign performance at an hourly level and adjust Schedule bid rules effectively, you can access the Report Center and download the campaign report ... Time period: Hourly ... This report is available for the last 30 days of your campaign. You can download 14 days of reporting data with a look-back period of 30 days." [official] *Bidding rules for Sponsored Products*.
+- GUIDANCE: "Apply a schedule-based budget rule to ensure that your campaign doesn’t exceed your budget due to bid increases." [official] same page. Budget rules are `areas/budgets-and-pacing.md`'s.
+- [seat] Amazon's native surface can raise bids for an hour window, a weekday or an event, and only upward, by 1–100% on the base bid. It cannot lower bids at night or pause a campaign by the hour. Third-party "dayparting" tools that pause or cut bids by the hour do so by writing bids or states through the API on a clock; Amazon's guide says a schedule rule change takes "within 24 hours", which says nothing about how fast a direct bid write takes effect.
+
+### I. Bid recommendations: suggested bid, range, themes, impact metrics
+
+- RULE: how Amazon says it computes them. "The suggested bid and bid range provided are calculated based on bids for ads similar to yours that have won impressions over the past 7 days. They're updated daily, based on the increase or decrease in competing bids and ads in each auction." [official] *Understand bidding on Sponsored ads* (help GDS2E64R9GQ35768). The API quickstart says recommendations are "calculated from a group of recent winning bids for similar products, updated periodically". [official] *Theme-based bid suggestions - Quick-start guide*, Ads API guide, https://advertising.amazon.com/API/docs/en-us/guides/sponsored-products/bid-suggestions/theme-based-bid-suggestions-quickstart-guide (path as read through the cloudfront markdown host). For Sponsored Display, the API says bid recommendations are "derived from the last 7 days of winning auction bids"; the 2021 launch note says "a group of winning bids for recent, similar ads within your category" that "update daily". [official] Sponsored Display 3.0 OpenAPI; *Sponsored Display launches bid recommendations*, what's new, 2021-09-20, https://advertising.amazon.com/resources/whats-new/sponsored-display-bid-recommendations
+- GUIDANCE: "We suggest using the suggested bid if you're new to Amazon Ads. To make your bid more competitive and improve the chance for your ad to display, start with the suggested bid or higher in the range." Example: "$0.70 ($0.40 - $1.40)". Above the range: "your ads will be more competitive in auctions". Below the suggested bid: "we recommend still staying within the suggested range". [official] *Understand bidding on Sponsored ads*.
+- RULE: event bids. "If a special event (such as Prime Day or a major holiday) is coming up in the next 45 days, we’ll also provide a suggested bid and bid range to use leading up to that event." "Higher CPCs typically occur in the weeks leading up to and during major shopping events such as Prime Day, Black Friday, and Cyber Monday." [official] *Set bids in a Sponsored Products campaign* (help GTMXQWASBRZTHLF2). In the API the themes are `CONVERSION_OPPORTUNITIES`, `PRIME_DAY`, `FALL_PRIME_DEAL_EVENT` and `BFCM_HOLIDAY`. [official] Sponsored Products 3.0 OpenAPI.
+- RULE: the endpoint. `POST /sp/targets/bid/recommendations` with content types `application/vnd.spthemebasedbidrecommendation.v3+json`, `v4` and `v5`. Requests are `BIDS_FOR_EXISTING_AD_GROUP` (campaign and ad group ids) or `BIDS_FOR_NEW_AD_GROUP` (up to 50 ASINs, plus a bidding strategy and up to 3 placement adjustments). Each target returns at most 3 `bidValues`. v4 adds product targets (`PAT_ASIN`, `PAT_CATEGORY`, `PAT_CATEGORY_REFINEMENT`) in all marketplaces and `KEYWORD_GROUP` (US only), and "removes 'impact metrics' when returning each bid suggestion". v5 adds "estimated impressions for the suggested bid" and, with `includeAnalysis`, eight bid points, each with `estimatedImpressionLower`, `Avg` and `Upper` per placement (`ALL`, `PLACEMENT_TOP`, `PLACEMENT_REST_OF_SEARCH`, `PLACEMENT_PRODUCT_PAGE`) and typed `SUGGESTED_LOWER`, `SUGGESTED`, `SUGGESTED_UPPER` or `ALTERNATIVE`; this is "currently available only for existing ad group requests in US". The endpoint "returns a 422 response when an unsupported marketplace or target is provided". A multi-country variant is `/sp/global/targets/bid/recommendations`. [official] Sponsored Products 3.0 OpenAPI.
+- RULE: the v3 content type is shut off. Amazon announced on 2024-08-22 the deprecation of `application/vnd.spthemebasedbidrecommendation.v3+json` "with a shutoff date of May 15, 2025" (callers to use v4), and on 2024-08-08 the shut-off of `/v2/sp/targets/bidRecommendations` on the same date; the deprecations page lists both. [official] *Amazon Ads advanced tools release notes*; *Deprecations*, https://advertising.amazon.com/API/docs/en-us/release-notes/deprecations . The OpenAPI still lists the v3 content type.
+- RULE (historical, v3-era): the quick-start guide describes impact metrics of weekly clicks and purchase orders for similar products that "are not estimates of expected performance", are "computed assuming your campaign does not run out of budget", and, for keyword targeting, need "a minimum number of five keywords per ad group" (no minimum for automatic targeting). [official] *Theme-based bid suggestions - Quick-start guide*.
+  - [seat] Those clicks-and-orders impact metrics belong to the v3 content type that was shut off on 2025-05-15; v4 removed them and v5 returns estimated impressions instead. The guide has not been updated to say so.
+- RULE: Performance Estimate (beta, not available to everyone) shows "the estimated weekly impressions a target may receive" across a range of bids. "For existing targets, Amazon uses historical advertising trends to estimate the number of impressions an ad may win at different bid values for a targeting clause. This estimation is based on recent trends and accounts for other advertisers’ bids and shopper engagement for each placement group." "It doesn’t aim to forecast future campaign performance". "Performance estimates don’t consider campaign budget constraints." [official] *Performance Estimate (beta)* (help GHFYQ2D4LQC5ABPW).
+- RULE: the unified API's bid recommendation. A `BID` recommendation type with scope `EXISTING_AD_GROUP`, `NEW_AD_GROUP` or `KEYWORD_AND_TARGET`, taking bidding settings, cost type, KPI, site restrictions, up to 100 targets and `includeAnalysis`; it returns a `BidValueRange` (`lower`, `suggested`, `upper`) and `BidImpactMetrics` (ranges of clicks, impressions and orders). [official] Amazon Ads API v1 merged OpenAPI. The release notes of 2026-08-31 announce `GET /adsApi/v1/list/recommendations` and `POST /adsApi/v1/query/recommendations`. [official] Ads API release notes.
+- RULE: campaign recommendations. `GET` and `POST /sp/campaign/recommendations` ("refreshed everyday") include a placement-bidding recommendation (action `ADD`, `DECREASE`, `INCREASE` or `REMOVE`, a suggested adjustment and an estimate of incremental impressions), a bidding-strategy recommendation, a shopper-cohort bidding recommendation and a targeting-group bid recommendation. The shopper-cohort recommendation describes its percentage as "basis points, e.g., 900 = 9%". [official] Sponsored Products 3.0 OpenAPI.
+  - [seat] The campaign field the recommendation feeds (`shopperCohortBidding.percentage`, 0–900) is a plain percent. Read literally, a recommended "900" means 9% in the recommendation and 900% in the campaign. Open question.
+- RULE: the Recommendations specification (`Recommendations_prod_3p.json`) lists recommendation types that touch bids: `AD_GROUP_BID_OPTIMIZATION`, `AD_GROUP_DEFAULT_BID`, `AMAZON_BUSINESS_BID_BOOST`, `AUDIENCE_COHORT_BID_BOOST`, `AUDIENCE_TARGETING_BID`, `CAMPAIGN_BIDDING_RULE`, `CAMPAIGN_BIDDING_STRATEGY`, `CAMPAIGN_PRODUCT_PLACEMENT`, `CAMPAIGN_TOP_PLACEMENT`, `KEYWORD_BID`, `NEW_CAMPAIGN_BIDDING_RULE` and `PRODUCT_TARGETING_BID`, with grouping types such as `INCREASE_BID_CONTEXTUAL`, `DECREASE_BID_CONTEXTUAL` and `OPTIMIZE_ROAS`. [official] Recommendations OpenAPI, https://dtrnk0o2zy01c.cloudfront.net/openapi/en-us/dest/Recommendations_prod_3p.json
+- RULE: Sponsored Brands bid recommendations "are built from historical winning bids for each specific target (keyword or product) and the impression opportunity for each specific goal": CPC recommendations for "Drive page visits", vCPM recommendations for "Grow brand impression share". [official] *Campaign goals for Sponsored Brands* (help GXXPFXLSVX9VD3U9), https://advertising.amazon.com/help/GXXPFXLSVX9VD3U9
+- RULE: Sponsored Display, `/sd/targets/bid/recommendations`: version 3.2 takes the optimization type and cost type; up to 100 targeting clauses. With cost control on, "Suggested bids aren’t available". [official] Sponsored Display 3.0 OpenAPI; *Understand Cost controls for sponsored ads* (help GNJDAT6AEYD3XJA8), https://advertising.amazon.com/help/GNJDAT6AEYD3XJA8
+
+### J. Top-of-search impression share and the placement reports
+
+- RULE: the metric. "The percentage of top-of-search impressions earned out of all the top-of-search impressions that were eligible ... Various factors determine the eligibility for an impression including campaign status and targeting status." [official] *Reporting columns (version 3)*, Ads API reporting v3 guide, https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/columns . The 2021 launch: "the percentage of top-of-search impressions your campaign received out of the total top-of-search impressions it was eligible to serve on ... available for Sponsored Brands and Sponsored Products campaigns with a 90-day lookback window. For Sponsored Products, this metric gives the impression share for top-of-search (first page)." [official] *Top-of-search impression share metric*, what's new, 2021-09-22.
+- RULE: Sponsored Brands' goal page defines it per reserved keyword: "the amount of impressions the advertiser received for the reserved keyword divided by the total amount of impressions across all ads for the reserved keyword" (Kitchen Smart: 600 of 1,000 = 60%). [official] *Campaign goals for Sponsored Brands* (help GXXPFXLSVX9VD3U9).
+  - [seat] The two definitions differ in the denominator: "eligible" impressions (the reporting definition) against "all ads for the keyword" (the Sponsored Brands goal page). A share computed one way is not comparable with the other.
+- RULE: where it can be read in v3. `topOfSearchImpressionShare` is listed for `spCampaigns`, `sbCampaigns` and `sbTargeting` on the columns page; the Sponsored Products targeting report page lists it among `spTargeting` base metrics. Re-read 2026-11-09 for the gap fill: the columns page's entry reads "**Report types**: spCampaigns, sbCampaigns, sbTargeting" (no `spTargeting`); the *Targeting reports* page lists it in the base metrics of both the Sponsored Products and the Sponsored Brands sections; the *Campaign reports* page lists it in the Sponsored Products "Group by `campaign`" and "Group by `campaignPlacement`" additional metrics and in the Sponsored Brands base metrics and "Group by `campaign`". So the four report types `spCampaigns`, `spTargeting`, `sbCampaigns` and `sbTargeting` each carry it on at least one Amazon page, and no single page lists all four. `spCampaigns` also accepts the `campaignPlacement` grouping, which gives `placementClassification` ("The page location where an ad appeared") and the `campaignBiddingStrategy` column; `keywordBid` is in `spTargeting` and `spSearchTerm`. Retention is 95 days for Sponsored Products, 60 for Sponsored Brands and 65 for Sponsored Display, and a request spans at most 31 days. [official] *Campaign reports*, *Targeting reports* and *Reporting columns (version 3)*, Ads API reporting v3 guides, https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/report-types/campaign and https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/report-types/targeting
+- RULE: "For Sponsored Products, there is not a separate placement report": placement is a grouping of the campaign report. Sponsored Brands has `sbCampaignPlacement` (preview; not for campaigns outside the multi-ad-group model). [official] *Placement reports*, Ads API reporting v3 guide, https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/report-types/placement
+- RULE: the console's downloadable reports include "Campaign placement" ("a summary of your campaign's performance by placement (top of search, other placements)"), "Keyword Placement" ("targeting performance broken out by ad placement") and the "Search Term Impression Share" report ("for each search term, the percentage share and numeric rank of your impression share of the total impressions relative to other advertisers"). [official] *Downloadable reports in sponsored ads* (help GBYSPTSLR337JMLH), https://advertising.amazon.com/help/GBYSPTSLR337JMLH
+- RULE: in the unified (v1) reporting API, `metric.topOfSearchImpressionShare` is compatible with campaign, ad product, account and time dimensions and incompatible with target, targeting, search term, ad group, hour and placement classification. `metric.impressionShare` and `metric.impressionShareRank` need `searchTerm.value` and are account-wide. `placementClassification.value` is "The classification of ad placements into categories like top of search, gateway page, etc."; `hour.value` is available for Sponsored Products, Sponsored Brands, Sponsored Display, Sponsored TV and DSP. [official] *Search (Impression share)* metric page and *Placement classification (Supply)* dimension page, Ads API v1 reporting guides, https://advertising.amazon.com/API/docs/en-us/guides/reporting/ads-v1/overview (pages read through the cloudfront markdown host).
+  - [seat] So moving from v3 to v1 loses top-of-search impression share at the keyword or target level and in combination with placement; v1 has it per campaign only, and the Stream v2 `ads-performance-v1` dataset page lists no impression-share field. v3 reporting (with Marketing Stream Version 1 and Sponsored Ads v2 reporting) "will stop delivering data on June 30, 2027" [official] *Deprecations* page, section "September 2026 - Deprecation of Amazon Marketing Stream Version 1, v3 reporting, and Sponsored Ads v2 reporting".
+- RULE: v1 data retention: `hour.value` 14 days (maximum range 14 days); `date.value` 24 months (maximum range 120 days); `day.value` and `dayOfWeek.value` 24 months (15 months); `dateRange.value` 6 years. "By default, performance is measured and data is aggregated based on each campaign timezone" (readable as `timeZone.value`); "Future releases will include support for UTC time zone normalization". [official] *Dates and historical data retention*, Ads API v1 reporting guide, https://advertising.amazon.com/API/docs/en-us/guides/reporting/ads-v1/dates
+
+### K. Sponsored Brands bidding
+
+- RULE: placements. "In Sponsored Brands campaigns, you can only adjust bids for top of search and rest of search placements." [official] *Understand bid adjustments in Sponsored ads* (help GQ4R4FR4H56JL6V8). The Sponsored Brands help page lets the console adjust "Placements other than top of search", by an increase or a decrease: "If you bid $1 for a keyword and set a 50% bid adjustment decrease, then your bid for placements other than top of search may decrease to $.50". [official] *Sponsored Brands bids by placement and audiences* (help GC8K522SMNCPH99Z), https://advertising.amazon.com/help/GC8K522SMNCPH99Z
+- RULE: automated bidding. "Automated bidding: Use the toggle to allow Amazon to automatically optimize bids for placements below top of search based on conversion rates." [official] *Create a Sponsored Brands campaign* (help GF86HBCNDJUAC5WN), https://advertising.amazon.com/help/GF86HBCNDJUAC5WN . In the API (v4) this is `bidding.bidOptimization`: "If set to true, Amazon will automatically set the right placement adjustment and the bidAdjustmentsByPlacement field is ignored"; turning it from false to true resets the placement adjustments to null. `bidAdjustmentsByPlacement` takes at most 3 entries from −99 to 900 over the placements `HOME`, `DETAIL_PAGE`, `OTHER` and `TOP_OF_SEARCH`; `shopperCohortBidAdjustments` takes at most 1. [official] Sponsored Brands 4.0 OpenAPI, https://d3a0d0y2hgofx6.cloudfront.net/openapi/en-us/sponsored-brands/4-0/openapi.json . The older v3 model: `bidOptimization` "Set to `true` to allow Amazon to automatically optimize bids for placements below top of search", default `true`; `bidMultiplier` from −99.00 to 99.00, settable only with `bidOptimization` false ("If set to -40.00 for a $5.00 bid, the resulting bid is $3.00"). [official] Sponsored Brands 3.0 OpenAPI, https://d3a0d0y2hgofx6.cloudfront.net/openapi/en-us/sponsored-brands/3-0/openapi.yaml
+- RULE: range changes. The note dated 2025-06-19 says advertisers "can now apply bid adjustments up to 900% for the top search placements" of Sponsored Brands, and that rest-of-search adjustments, "previously limited to up to 99% can now also be adjusted up to 900%" (a later note dates the top-of-search introduction 6/17/2025). On 2025-12-01 Amazon announced that negative Sponsored Brands placement adjustments "will not be supported starting February 2nd, 2026"; on 2026-01-27 it extended that to 2026-06-15: "existing campaigns with negative bid adjustments will continue to run with their current settings, no new negative adjustments will be accepted after this date." On 2023-10-18 Amazon deprecated `bidOptimizationStrategy` and `bidAdjustmentsByShopperSegment` for 2024-03-31. [official] Ads API release notes.
+  - [seat] So the "decrease by" option on the console help page and the −99 floor in the v4 model both predate the 2026-06-15 cut-off; a write of a negative adjustment after that date is expected to be refused. The live model is the judge.
+- RULE: dynamic adjustment on Sponsored Brands. "In Sponsored Brands, if we find an opportunity where your ad is more likely to drive conversions, we may increase your bid for that auction up to 300%. Similarly, if we find an opportunity that looks less likely to drive conversions, we might lower your bid for that auction. For Sponsored Brands, these bid adjustments are available only on campaigns where you select Drive page visits as your campaign goal. For example, when bidding for ad placements, Amazon can adjust a bid of $1 up to $4". [official] *Understand bidding on Sponsored ads* (help GDS2E64R9GQ35768). The release notes date this to 2023-10-05 ("Drive Page Visits" CPC campaigns). [official] Ads API release notes.
+  - [seat] The v1 model's note that `placementBidAdjustments` are "Not supported for Sponsored Brands campaigns using the SALES_UP_AND_DOWN bid strategy" reads as the same switch: when Amazon optimizes Sponsored Brands bids up and down, placement adjustments are not taken.
+- RULE: audiences on Sponsored Brands: "You can adjust bids for only 1 audience per campaign. You can choose either an Amazon-built audience or an Amazon Marketing Cloud (AMC) audience, but not both." Up to 900%. Placement and audience multiply: $1, +50% placements other than top of search = $1.50, +100% custom audience = $3. [official] *Sponsored Brands bids by placement and audiences* (help GC8K522SMNCPH99Z). The API guide gives the same chain with a detail page adjustment: $2.00, `DETAIL_PAGE` +40% = $2.80, +50% audience = $4.20. [official] *Shopper cohort bidding*, Sponsored Brands API guide, https://advertising.amazon.com/API/docs/en-us/guides/sponsored-brands/bidding/shopper-cohort
+- RULE: cost control. "If your campaign goal is "drive page visits," cost control is automatically turned on." "Because cost control is a target average cost and not a max cost, your average CPC may be higher than the amount entered." "If you turn off cost control, you'll need to enter in a specific bid for keywords or products". "You can only use cost control if your campaign goal is Drive page visits." With placement or audience adjustments, "your highest possible CPC will exceed your target cost control CPC". After launch the target "will display on the campaign settings page for Sponsored Brands as a 7-day rolling average". [official] *Understand Cost controls for sponsored ads* (help GNJDAT6AEYD3XJA8); *Sponsored Brands bids by placement and audiences*. (The cost-control page's bullets "Bids can’t be manually edited" and "Suggested bids aren’t available" sit under its Display heading, see L.) The API's example: a $2.00 cost control with a +50% audience segment gives a $3.00 target. [official] *Shopper cohort bidding* (Sponsored Brands API guide).
+  - [seat] Cost control is the one bid mechanism on Amazon whose number is an average, not a ceiling. Every other page in this file repeats "never exceed the maximum bid"; for cost-controlled Sponsored Brands, individual clicks can cost more than the number the seller typed.
+- RULE: Sponsored Brands optimization rules (beta; the API form of cost control): for campaigns with goal `PAGE_VISIT` "you can apply a cost-based rule for clicks (CPC)", at `/sb/rules/optimization` with `/associate`, `/disassociate` and `/list`; a threshold recommendation from `/sb/recommendations/optimization` (metric `COST_PER_CLICK`, returning `recommendedValue` and `minimumValue`); "the system may add targets while your campaign is running"; "While Amazon Ads cannot guarantee performance, if the campaign is not meeting the cost target over a 21-day period (excluding special days), we recommend adjusting cost target or budget". [official] *Managing Sponsored Brands optimization rules*, Sponsored Brands API guide, https://advertising.amazon.com/API/docs/en-us/guides/sponsored-brands/campaigns/managing-optimization-rules . The rule's only attribute, `COST_PER_CLICK` with `LESS_THAN_OR_EQUAL_TO`, is described as "Maximize page visits while cost per click less than or equal to threshold", and the recommendation request takes 1 to 10 landing pages. [official] Sponsored Brands 4.0 OpenAPI.
+- RULE: goals and cost types. Goals are `PAGE_VISIT` (the default) and `BRAND_IMPRESSION_SHARE`; `ACQUIRE_NEW_CUSTOMERS` and `AD_VIEWS` are "PREVIEW ONLY"; cost types are `CPC` (default), `VCPM` and `FIXED_PRICE` ("It can only be used for campaign with a targetedPGDealId", the deal that reserve share of voice campaigns use); `smartDefault` is "not editable once the campaign is created"; "Goal-based campaigns are not supported in the Ireland (IE) marketplace" (supplying `goal` and `costType` there returns `GOAL_BASED_CAMPAIGNS_DISABLED`). [official] Sponsored Brands 4.0 OpenAPI. "Once a goal-based campaign is created, the `goal`, `costType`, and `smartDefault` values are not editable"; "For `goal` type `BRAND_IMPRESSION_SHARE`, only `costType` `VCPM` is available"; and the tip is not to include bidding optimizations (`bidOptimizationStrategy`, `bidAdjustmentsByShopperSegment`, `bidAdjustmentsByPlacement`, `bidOptimization`) when `smartDefault` is `TARGETING` or the cost type is `VCPM`. [official] *Getting started with campaigns*, Sponsored Brands API guide, https://advertising.amazon.com/API/docs/en-us/guides/sponsored-brands/campaigns/get-started-with-campaigns . "You can choose only 1 goal per campaign, and this goal can’t be changed after the campaign is created." [official] *Campaign goals for Sponsored Brands* (help GXXPFXLSVX9VD3U9).
+- RULE: reserve share of voice. "Instead of bid recommendations, Reserve brand share of voice (SOV) provides a fixed, upfront price for top of search visibility for your branded keywords without having to bid in auction. You can upload or enter 5 to 1,000 branded keywords ... Your pricing is for the duration of your campaign, which must be between 30 to 92 days." "We strive to deliver 95% SOV at the campaign level when campaigns are properly set up". "Reserve share of voice is not available in Spain." [official] *Campaign goals for Sponsored Brands*. The product page says it "requires a minimum spend commitment". [official] *Sponsored Brands*, Amazon Ads product page, https://advertising.amazon.com/solutions/products/sponsored-brands . In the API a proposed deal (`deal.state` `PROPOSED`) triggers pricing; the placements are Sponsored Brands top of search for branded keywords "at a fixed, upfront price ... most of the time"; Amazon announced the API on 2025-10-15. [official] *Reserve share of voice*, Sponsored Brands API guide, https://advertising.amazon.com/API/docs/en-us/guides/sponsored-brands/reserve-share-of-voice ; Ads API release notes. Its use as brand defence is `beyond/brand-defense-and-competition.md`'s.
+  - Amazon's reserve share of voice pages disagree on the flight, the minimum and the marketplaces (figures for human readers only; re-read 2026-11-09): *Campaign goals for Sponsored Brands* (help GXXPFXLSVX9VD3U9) says "between 30 to 92 days" and "not available in Spain"; *Reserve keywords in a Sponsored Brands campaign* (help G86SD7HK6NHHRB9B, https://advertising.amazon.com/help/G86SD7HK6NHHRB9B) says "the campaign duration must be between 15 and 92 days", "A minimum budget of $3,000", a start 3 days to 3 months out, keywords mapped to an ad within 14 days, and also "not available in Spain"; the library guide *How to set up Amazon Sponsored Brands reserve share of voice* (https://advertising.amazon.com/library/guides/sponsored-brands-reserve-share-of-voice) says "campaign flight (e.g., 14–90 days)", "a minimum spend of $6K per campaign for the U.S. marketplace" and lists 20 marketplaces including Ireland and excluding Spain. The full set of discrepancies, including the what's-new page's marketplace list, is `beyond/brand-defense-and-competition.md`'s (P-6, P-39). [official] pages as named, read 2026-11-09.
+- RULE: an Amazon-Business-only Sponsored Brands campaign cannot change its site restriction after creation and does not support shopper-cohort bidding. [official] Sponsored Brands 4.0 OpenAPI.
+
+### L. Sponsored Display (now "display ads"): bid optimization, vCPM, cost control
+
+- RULE: naming. "Sponsored Display is now known as display ads", part of the Amazon Ads Agent; existing campaigns keep running. [official] *Sponsored Display*, Amazon Ads product page, https://advertising.amazon.com/solutions/products/sponsored-display . The product page says existing Sponsored Display campaigns are reached "through Display, Video, and Audio campaigns in Amazon Ads Agent"; the help pages call the product "Display" and point to a separate article for "bidding in Display, Video, and Audio campaigns". [official] *Sponsored Display* product page; *Understand bidding on Sponsored ads* (help GDS2E64R9GQ35768).
+- RULE: the 300% adjustment. "If we find opportunities to increase viewable impressions, clicks, conversions, or leads based on the optimization strategy you select for your ad, we may increase your bid for that auction up to 300%. Similarly, if your ad looks less likely to achieve your optimization goal, we may lower your bid for that auction. Example: When bidding for ad placements, Amazon can adjust a bid of $1 up to $4". [official] *Optimization strategies for Display* (help GDRFG6U7TRD72EFW), https://advertising.amazon.com/help/GDRFG6U7TRD72EFW . The API overview: "Regardless of your bid optimization selection ... we may increase your bid for that auction by up to 300%". [official] *Amazon Ads API for Sponsored Display overview*, Ads API guide, https://advertising.amazon.com/API/docs/en-us/guides/sponsored-display/overview
+- RULE: the strategies and their pricing. `bidOptimization` `clicks` (CPC, the default, "Optimize for page visits"), `conversions` (CPC) and `reach` (vCPM); the help page adds "Optimize for leads" (CPC). "Because optimization strategies are set at the campaign level, once you select a strategy, you won’t be able to change it." [official] *Amazon Ads API for Sponsored Display overview* (API); *Optimization strategies for Display* (help). Amazon is moving reach and page-visit campaigns of sellers on Amazon toward the conversion objective: "These campaigns will now align with the optimization objective of Sponsored Display Conversion campaigns" (help); the release notes date the move 2025-04-04. [official]
+- RULE: vCPM. "when your sponsored ad gets 1,000 viewable impressions, your bid amount is deducted from your daily or lifetime budget"; a view is MRC's: "At least 50% of the ad should have been in the customer's viewport for at least 1 second" (video: 50% with 2 seconds of playback). [official] *Understand bidding on Sponsored ads*. On floors the pages disagree (for human readers only): the API guide suggests "starting with $5 USD bids" and says "$1 is the minimum bid for vCPM"; the help page warns "If you enter a vCPM bid of less than $3, your ads may not lead to viewable impressions". [official] *Amazon Ads API for Sponsored Display overview* (API); *Optimization strategies for Display* (help).
+- RULE: attribution differs by strategy. For sellers on Amazon, "attribution of optimize for viewable impressions is based on clicks and estimated views—similar to Amazon DSP. Other optimization strategies, as well as Sponsored Products and Sponsored Brands campaigns’ attribution types are based solely on clicks." [official] *Optimization strategies for Display*. Metric meanings are `beyond/metrics-and-attribution.md`'s.
+- RULE: display cost control. "When using cost control, you choose a target average cost per click (CPC)"; "Your bid is automatically set and adjusts periodically based on ad performance to try to meet your target CPC"; "Bids can’t be manually edited"; "Suggested bids aren’t available". [official] *Understand Cost controls for sponsored ads* (help GNJDAT6AEYD3XJA8). In the API (beta) the rule metric follows the strategy: reach → `COST_PER_THOUSAND_VIEWABLE_IMPRESSIONS`, clicks → `COST_PER_CLICK`, conversions → `COST_PER_ORDER`; "When an optimization rule is associated to an ad group, manual bids for individual targets will be overridden"; one rule per ad group; the campaign budget must be at least 5× the cost-per-order threshold, 10× the vCPM threshold and 20× the CPC threshold; "We may add targets while your campaign is running to try to stay at or below a cost per order value you have specified". Paths `/sd/optimizationRules` and `/sd/adGroups/{adGroupId}/optimizationRules`. [official] Sponsored Display 3.0 OpenAPI, https://d3a0d0y2hgofx6.cloudfront.net/openapi/en-us/sponsored-display/3-0/openapi.yaml . US minimum thresholds (for human readers only): `COST PER THOUSAND VIEWABLE IMPRESSIONS` 1, `COST PER CLICK` 0.5, `COST PER ORDER` 5. [official] *Limits, constraints, and quotas*.
+- RULE: Sponsored Display in the unified v1 campaign API (corrected in the gap fill of 2026-11-09). *Campaign management in the Amazon Ads API* (the overview) reads "The Amazon Ads API currently supports campaign management operations for:" Amazon DSP, Sponsored Brands, Sponsored Products, Sponsored Display and Sponsored Television. Amazon publishes a Sponsored Display v1 specification (`AmazonAdsAPISD_prod_3p.json`, title "Amazon Ads API SD 3.0") with `POST /adsApi/v1/update/campaigns`, `update/adGroups` and `update/ads` (at most 100 items each) and `update/targets` (at most 1,000). In it an SD ad group's bid is `bid.defaultBid` ("The default maximum bid for ads and targets in the ad group") and a target's is `bid.bid`; the ad group's `optimization.goalSettings.kpi` takes `ADD_TO_CART`, `APPLICATIONS`, `CHECKOUTS`, `CLICKS`, `CONTACTS`, `LEADS`, `OTHER`, `PAGE_VIEWS`, `PURCHASES`, `REACH`, `SEARCH`, `SIGN_UP` or `SUBSCRIBE`; the SD campaign update carries no `bidSettings` and its `costType` is `CPC` or `VCPM`. Against this: the release notes of 2025-08-01 and 2025-12-01 promised Sponsored Display "later in 2025" / "later this year", no release note read on 2026-11-09 dates SD general availability on v1, and the campaign entity guide's field table still has columns only for the common model, Sponsored Products, Sponsored Brands and Amazon DSP. [official] *Campaign management in the Amazon Ads API*, https://advertising.amazon.com/API/docs/en-us/guides/campaign-management/overview ; *Amazon Ads API v1, Sponsored Display* model, https://d1y2lf8k3vrkfu.cloudfront.net/openapi/en-us/dest/AmazonAdsAPISD_prod_3p.json ; *Release notes: Ads API v1*, https://advertising.amazon.com/API/docs/en-us/release-notes/ads-api-v1 ; *The campaign entity in the Amazon Ads API*, https://advertising.amazon.com/API/docs/en-us/guides/campaign-management/entities/campaign ; all read 2026-11-09.
+  - Developers report SD v1 behaviour that departs from the model (an SD v1 budget update silently ignored, issue #475; an SD v1 KPI refused, issue #481); those field reports and the settlement of what is usable belong to `method/ads-api-surface.md` (§E, §J). [opinion] GitHub field reports as recorded there.
+  - [seat] The earlier wording here ("not yet in the unified v1 campaign API") was wrong on Amazon's current overview and published model. What is true is narrower: Amazon documents SD v1 operations and lists SD as supported, has not dated its general availability in a release note, and developers report SD v1 writes that do not take effect as written.
+
+### M. How soon bid-relevant data arrives
+
+- RULE: "Impression/click event to availability via API" 12 hours; "Impression/click invalidation" 72 hours. [official] *Limits, constraints, and quotas*.
+- RULE: hourly Sponsored Products performance in Amazon Marketing Stream carries `keyword_id`, `match_type`, `placement` ("The page location where an ad appeared") and `time_window_start` in the advertiser's time zone. "Initial click data is available within 12 hours of the ad click. As part of the traffic validation process, clicks can be invalidated during the following 72 hours post the initial report." Conversions "are reported based on the hour when the click they are attributed to occurred ... you should expect to receive revisions to conversion data up to sixty days after the initial click." Data "will be delivered to your queue in a burst every hour". [official] *Sponsored Products performance datasets*, Amazon Marketing Stream guide, https://advertising.amazon.com/API/docs/en-us/guides/amazon-marketing-stream/datasets/sp-performance . The page describes two Stream Version 1 datasets, and "sp-performance" is only the page's path: "To subscribe to this dataset, use the ID `sp-traffic`" (clicks, impressions, cost; the 12-hour and 72-hour freshness lines) and "use the ID `sp-conversion`" (conversions; the sixty-day revision line). Both carry `keyword_id`, `placement` and `time_window_start`. Stream Version 1 "will stop delivering data on June 30, 2027". [official] *Sponsored Products performance datasets* (re-read 2026-11-09); *Deprecations*.
+- RULE: Stream v2 delivers Parquet files to the subscriber's own S3 bucket "As frequently as every 15 minutes"; "Each record carries the total value for its time window, not a change since the last delivery. Stream v2 delivers only the records that changed; when a record is restated, you receive it again with its new total." Its `ads-performance-v1` dataset has an hourly grain and a "Restatement window" of 70 days, and carries `placementClassification.value` and `hour.value` for sponsored ads. Stream v2 and the Reporting API v1 became generally available on 2026-09-30. [official] [official] *Overview* (Amazon Marketing Stream Version 2), https://advertising.amazon.com/API/docs/en-us/guides/amazon-marketing-stream/v2/overview ; *Ads performance* (Stream v2 dataset page), https://advertising.amazon.com/API/docs/en-us/guides/amazon-marketing-stream/v2/datasets/ads-performance
+- RULE: "Daily budgets are not paced throughout the day". [official] *Sponsored Products*, Amazon Ads product page, https://advertising.amazon.com/solutions/products/sponsored-products . (Pacing is `areas/budgets-and-pacing.md`'s; it is recorded here because an hour's CPC depends on which competitors still have budget, see Perpetua below.)
+- [seat] Taken together: a bid change's clicks are visible after up to 12 hours, can be withdrawn for 72 hours more, and its conversions can be restated for 60 days. A read of "what this bid did" taken the next morning is provisional on all three counts. `method/change-measurement.md` owns how to read a change.
+
+### N. Naming, surfaces and migration in 2025–2026
+
+- RULE: the unified campaign API (v1) reduces "nearly 200" endpoint variations "to 16" (2025-12-01); bid settings live under `bidSettings` with `bidStrategy` and `bidAdjustments` for Sponsored Products, Sponsored Brands and DSP. [official] Ads API release notes; *The campaign entity in the Amazon Ads API*.
+- RULE: the Amazon Ads MCP server offers "AI migration skills" (announced 2026-11-02), starting with `ads-api-migration` for campaign management, "providing mappings of APIs, fields, and enums"; skills download from the MCP server's `/skills` path on the Ads API hosts. [official] *AI Migration Skills*, https://advertising.amazon.com/API/docs/en-us/mcp/ai-migration-skills ; *Download Skills from the Amazon Ads MCP Server*, https://advertising.amazon.com/API/docs/en-us/mcp/skills/get-started . (The skills themselves sit on an Ads API host and were not fetched; see *Unreachable*.)
+- RULE: the Sponsored Products 3.0 campaign model carries an `autoManageCampaign` boolean with no description. [official] Sponsored Products 3.0 OpenAPI. What it switches is `beyond/automation-and-amazon-tools.md`'s question.
+
+### O. The bid writes: operations, bodies, batch ceilings and per-item refusals
+
+Every item below is RULE, read from the OpenAPI model named at its end on 2026-11-09 (models fetched from their public cloudfront addresses; no Ads API host called). Batch ceilings are figures **for human readers only**; the model a call is made against is the authority. Each write carries the usual `Amazon-Advertising-API-ClientId` and `Amazon-Advertising-API-Scope` headers (the `cx raw amazon-ads` door supplies the token). The call mechanics (throttling, retries, the 207 contract in general) are `method/ads-api-surface.md`'s.
+
+Sponsored Products, version 3 (`SponsoredProducts_prod_3p.json`):
+
+- `PUT /sp/keywords` (operation `UpdateSponsoredProductsKeywords`), media type `application/vnd.spKeyword.v3+json` on request and 207 response. Body `{"keywords": [ … ]}`, at most 1,000 items; each item requires `keywordId` and may carry `bid` (double, nullable: "Bid associated with this keyword. Applicable to biddable match types only") and `state`.
+- `PUT /sp/targets` (`UpdateSponsoredProductsTargetingClauses`), `application/vnd.spTargetingClause.v3+json`. Body `{"targetingClauses": [ … ]}`, at most 1,000; each requires `targetId` and may carry `bid` (nullable: "Targets that do not have bid values in listTargetingClauses will inherit the defaultBid from the adGroup level"), `state`, `expression` and `expressionType` (`AUTO` or `MANUAL`). The automatic-targeting clauses of an auto campaign are targets of `expressionType` `AUTO` and take a `bid` here.
+- `PUT /sp/adGroups` (`UpdateSponsoredProductsAdGroups`), `application/vnd.spAdGroup.v3+json`. Body `{"adGroups": [ … ]}`, at most 1,000; each requires `adGroupId` and may carry `defaultBid` ("A bid value for use when no bid is specified for keywords in the ad group"), `name` and `state`.
+- `PUT /sp/campaigns` (`UpdateSponsoredProductsCampaigns`), `application/vnd.spCampaign.v3+json`. Body `{"campaigns": [ … ]}`, at most 1,000 ("Note: targetingType cannot be updated"); each requires `campaignId`. The bid fields sit in `dynamicBidding`: `strategy` (`LEGACY_FOR_SALES`, `AUTO_FOR_SALES`, `MANUAL`, `RULE_BASED` on writes; `OTHER` appears only on reads), `placementBidding[]` of `{placement, percentage}` (percentage int32 0–900; placements `PLACEMENT_TOP`, `PLACEMENT_REST_OF_SEARCH`, `PLACEMENT_PRODUCT_PAGE`, `SITE_AMAZON_BUSINESS`) and `shopperCohortBidding[]` (at most 10). `offAmazonSettings` is on the same update; `siteRestrictions` appears in the update shape but "cannot be changed post campaign creation" (D).
+- The optional `Prefer: return=representation` header asks the service to "return the full object when doing create/update/delete operations instead of ids".
+- The 207 body is `{"success": [ {index, keywordId|targetId|…, <entity>} ], "error": [ {index, errors: [ … ]} ]}`, where `index` is "the index of the keyword in the array from the request body". Each error carries one typed member. The bid member, `biddingError`, appears in the keyword, target, ad group and campaign error selectors with `reason`, `message`, `marketplace`, `lowerLimit` and `upperLimit`; its reasons are `BID_OUT_OF_MARKET_PLACE_RANGE`, `BID_GT_BUDGET`, `BID_INVALID_PLACEMENT`, `BID_AUDIENCES_MORE_THAN_ALLOWED`, `BID_INVALID_AUDIENCE_ID`, `BID_INVALID_AUDIENCE_SEGMENT_TYPE`, `BID_INVALID_SHOPPER_COHORT_TYPE`, `BID_MISSING_AUDIENCES` and `BID_SHOPPER_COHORTS_MORE_THAN_ALLOWED`. The state member, `entityStateError`, has reasons including `ARCHIVED_ENTITY_CANNOT_BE_MODIFIED`, `PARENT_ARCHIVED_FORBIDS_UPDATES`, `PARENT_STATUS_FORBIDS_UPDATES_AND_CREATES`, `INVALID_STATE_TRANSITION` and `AUTO_TARGETING_CLAUSE_CANNOT_BE_ARCHIVED_MANUALLY`. A campaign budget lowered below the bids beneath it is refused by `budgetError` reasons `BUDGET_LT_DEFAULT_BIDS`, `BUDGET_LT_KEYWORD_BIDS` and `BUDGET_LT_PREDEFINED_TARGET_BIDS`. Whole-request failures are 400 (`KeywordMutationException` and the like), 415 for a wrong media type, and 429 for throttling.
+[official] Sponsored Products 3.0 OpenAPI, https://d1y2lf8k3vrkfu.cloudfront.net/openapi/en-us/dest/SponsoredProducts_prod_3p.json
+
+Unified v1 (`AmazonAdsAPIALLMerged_prod_3p.json` and the per-product models):
+
+- Every v1 write is a `POST` with media type `application/json`: `POST /adsApi/v1/update/targets` (`UpdateTarget`; body `{"targets": [ {targetId, bid: {bid} | {marketplaceSettings[]}, state, …} ]}`), `POST /adsApi/v1/update/adGroups` (`UpdateAdGroup`; `bid.defaultBid`, "The default maximum bid for ads and targets in the ad group. This is used in sponsored ads as the maximum bid during the auction"), and `POST /adsApi/v1/update/campaigns` (`UpdateCampaign`; `optimizations.bidSettings.bidStrategy` and `optimizations.bidSettings.bidAdjustments.{placementBidAdjustments, audienceBidAdjustments, creativeBidAdjustments, shopperSegmentBidAdjustments}`). A keyword is a target in v1. The merged model sets each request array to at most 1,000 and notes "Batch size limits are specific to each ad product".
+- The per-product models give the ceilings (for human readers only): Sponsored Products 1,000 for campaigns, ad groups and targets; Sponsored Brands 10 for campaigns and ad groups and 1,000 for targets; Sponsored Display 100 for campaigns and ad groups and 1,000 for targets.
+- The per-product models also narrow the bid fields. Sponsored Products v1: `bidStrategy` `MANUAL`, `RULE_BASED`, `SALES_DOWN_ONLY` or `SALES_UP_AND_DOWN`; `placementBidAdjustments` at most 4 over `TOP_OF_SEARCH`, `REST_OF_SEARCH`, `PRODUCT_PAGE`, `SITE_AMAZON_BUSINESS`; `audienceBidAdjustments` at most 1 (the merged model says 10); `creativeBidAdjustments` at most 2. Sponsored Brands v1: `bidStrategy` `MANUAL` or `SALES_UP_AND_DOWN`; placements `TOP_OF_SEARCH`, `REST_OF_SEARCH`, `PRODUCT_PAGE`, `HOME_PAGE`, at most 4; one audience; ad groups carry no bid (Sponsored Brands bids are on targets). Sponsored Display v1: see L.
+- The 207 body is `{"success": [ {index, target} ], "partialSuccess": [ … ], "error": [ {index, errors: [ {code, message, fieldLocation} ]} ]}`. The codes that refuse a bid write include `FIELD_VALUE_IS_OUT_OF_RANGE`, `FIELD_VALUE_IS_ABOVE_MAXIMUM_LIMIT`, `FIELD_VALUE_IS_BELOW_MINIMUM_LIMIT`, `ARCHIVED_RESOURCE_CANNOT_EDIT` ("Archived resources cannot be edited"), `ARCHIVED_PARENT_CANNOT_EDIT` ("Resources within an archived parent cannot be edited") and `AUTOCREATED_ENTITY_CANNOT_EDIT` ("Autocreated entities cannot be edited. To complete this action, create the resource manually"). v1 has no bid-specific error code named like v3's `BID_GT_BUDGET`. The model's `ErrorsIndex.index` is bounded 0–19 while success indexes run to 999.
+[official] Amazon Ads API v1 merged OpenAPI, https://d1y2lf8k3vrkfu.cloudfront.net/openapi/en-us/dest/AmazonAdsAPIALLMerged_prod_3p.json ; *Amazon Ads API v1, Sponsored Products / Sponsored Brands / Sponsored Display* models (`AmazonAdsAPISP_prod_3p.json`, `AmazonAdsAPISB_prod_3p.json`, `AmazonAdsAPISD_prod_3p.json`) at the same path.
+  - [seat] An error index capped at 19 in a 1,000-item batch reads as a model defect or a sign that errors past the twentieth item are not itemised; the model does not say which. A read-back of the written entities settles what happened.
+
+Sponsored Brands (versions 3 and 4):
+
+- `PUT /sb/keywords` (`updateKeywords`, SB 3.0 model): request `application/json`, a bare array of keyword objects ("Operation supports a maximum list size of 100 keywords"), each requiring `keywordId`, `adGroupId` and `campaignId`, with `bid` ("must be less than the budget associated with the Advertiser account") and `state`. The request body note: "the bid is only mutable when the keyword's corresponding campaign does not have any enabled optimization rule." The 207 response (`application/vnd.sbkeywordresponse.v3+json`) is an array, in request order, of `{keywordId, code, description, errors}`.
+- `PUT /sb/targets` (`updateTargets`): request `application/json`, body `{"targets": [ … ]}` with 1 to 100 items, each with `targetId`, `adGroupId`, `campaignId`, `state` and `bid`. Request note: targets "can be updated on campaigns where serving status is not one of `archived`, `terminated`, `rejected`, or `ended`" and the same optimization-rule sentence. Unlike the other writes it answers 200 (`application/vnd.updatetargetsresponse.v3+json`) with `updateTargetSuccessResults` and `updateTargetErrorResults`, each keyed by `targetRequestIndex`.
+- `PUT /sb/v4/campaigns` (SB 4.0 model): media type `application/vnd.sbcampaignresource.v4+json`, body `{"campaigns": [ … ]}` with 1 to 10 items, each requiring `campaignId`. Bid fields sit in `bidding`: `bidOptimization` (true: Amazon "will automatically set the right placement adjustment and the bidAdjustmentsByPlacement field is ignored"), `bidAdjustmentsByPlacement` (at most 3 `{placement, percentage}`, −99 to 900, "can only be set when 'bidOptimization' is set to false") and `shopperCohortBidAdjustments` (at most 1). `PUT /sb/v4/adGroups` carries only `name` and `state`. The 207 body is `{"campaigns": {"success": [ … ], "error": [ {index, errors: [ {errorType, errorValue} ]} ]}}`, where `errorValue` holds a `biddingError`, `budgetError`, `rangeError`, `dateError`, `billingError` or `otherError`, each with a free-text `reason`, `message`, a `cause.location` JSON path and, for bid, budget and range errors, `lowerLimit` and `upperLimit`.
+[official] Sponsored Brands 3.0 OpenAPI, https://d3a0d0y2hgofx6.cloudfront.net/openapi/en-us/sponsored-brands/3-0/openapi.yaml ; Sponsored Brands 4.0 OpenAPI, https://d3a0d0y2hgofx6.cloudfront.net/openapi/en-us/sponsored-brands/4-0/openapi.json
+  - [seat] The −99 floor in the v4 shape predates the 2026-06-15 end of new negative Sponsored Brands placement adjustments (K); the model was not updated.
+
+Sponsored Display, version 3 (`sponsored-display/3-0/openapi.yaml`):
+
+- `PUT /sd/targets` (`updateTargetingClauses`): request `application/json`, a bare array of `{targetId, bid, state, …}`. The model states the ceiling twice and differently: "Maximum length of the array is 100 objects" (operation description) and "A list of up to 1000 targeting clauses" (request body). The mutable fields are `state` and "`bid` (only mutable when the targeting clause's adGroup does not have any enabled optimization rule)". The target `bid` "will override the adGroup bid if specified"; "$1 is the minimum bid for vCPM".
+- `PUT /sd/adGroups` (`updateAdGroups`): a bare array, "Maximum length of the array is 100 objects"; "The mutable fields are 'name', 'defaultBid', and 'state'". `bidOptimization` (`reach`, `clicks`, `conversions`) is not among them.
+- Both answer 207 with an array of `{code, description, targetId|adGroupId}` per item, and 422 for "Unprocessable Entity - Request was understood, but contained invalid parameters".
+- Cost control and optimization rules: "When an optimization rule is associated to an ad group, manual bids for individual targets will be overridden" (the optimization-rule operations), and the console help says of display cost control "Bids can’t be manually edited" (L).
+[official] Sponsored Display 3.0 OpenAPI, https://d3a0d0y2hgofx6.cloudfront.net/openapi/en-us/sponsored-display/3-0/openapi.yaml ; *Understand Cost controls for sponsored ads* (help GNJDAT6AEYD3XJA8).
+  - [seat] Amazon's two wordings ("only mutable when … does not have any enabled optimization rule" and "will be overridden") do not say whether a bid write to such a target is refused with an item error or accepted and then overridden. Either way the written bid does not govern the auction while the rule is on; a read-back of the target and of the ad group's rules tells which happened. The same holds for Sponsored Brands keywords and targets under an enabled optimization rule.
+
+## What practitioners add
+
+Each item is [opinion] unless marked. Practitioners and vendors sell bid tools; their data are observational unless they say otherwise.
+
+### On the auction and what an advertiser pays
+
+- Ad Badger (The Badger), on the FTC case: the FTC's figure (advertisers paying their own bid about 80% of the time in 2024) is about price paid, while Amazon's (92% of winners were not the highest bidder) is about who wins; "both sets of numbers can be true": "'Not the highest bid' is about who wins ... 'Paying your own bid' is about what you are charged after you win." It suggests that ratios of CPC to bid "clustering at 0.90–1.00 mean you are effectively in a first-price auction". *Amazon Says Its Ad Auction Was Second-Price. The FTC Says You Paid Your Own Bid Anyway.*, Ad Badger blog, 2026-09-10, https://www.adbadger.com/blog/amazon-says-its-ad-auction-was-second-price-the-ftc-says-you-paid-your-own-bid-anyway/ (fetched by the verifier on 2026-11-09).
+- The "second price plus one cent" rule. Ad Badger's FTC post describes the historical pitch as "if you win, you pay one cent more than the next-highest bid"; vendor pages still teach it (seen in search-result summaries of vendor guides, 2026-11-09). [seat] Amazon's own description (A) makes this out of date: the soft reserve can lift the price above the runner-up, up to the winner's maximum.
+- Ad Badger (The Badger, 2018-02-21) teaches the bid as revenue per click times target ACOS: "average order value times your conversion rate, divided by one over your Target ACoS"; $19 × 10% = $1.90 revenue per click, × 30% = $0.57. The article treats bid and cost per click as the same number. *Calculating a Bid Strategy for Amazon Ad Campaigns*, https://www.adbadger.com/blog/calculating-amazon-bid-strategy/
+- Trellis (Geoffrey Martlin) gives the same ceiling per placement: "maximum CPC = price x conversion rate in that placement x target ACoS", and the placement adjustment ceiling as maxCPC ÷ base bid − 1 (its illustration: a $30 product, 30% target ACOS and $0.60 base bid; at 10% top-of-search conversion, $30 × 0.10 × 0.30 = $0.90, a +50% ceiling; at 14%, +110%; at 8%, +20%; at 5%, no ceiling because $0.45 is below the base bid). It advises reading the placement report first, moving one placement at a time, and treating the computed number "as a ceiling, not a target". It states the up-and-down ceiling as 100% for top of search and 50% elsewhere "according to a third-party summary of Amazon's announcement". *Amazon Placement Bid Adjustments*, Trellis blog, 2025-05-13, updated 2026-11-04, https://gotrellis.com/resources/blog/amazon-ad-placements
+  - [seat] Both formulas compute a maximum the seller is willing to pay per click. Under Amazon's auction (A) the bid is a maximum, and with soft reserves the price paid has drifted toward it (FTC figures), so the formula's output is closer to a cost estimate now than it was under a pure second-price rule. Economics of target ACOS belong to `beyond/goals-and-economics.md`.
+- On the suggested bid: Amalyze calls Amazon's figures "a backwards-looking summary of winning bids on auctions Amazon judged comparable" and states "The bid range is the 25th–75th percentile of recent winning bids ... The suggested bid is roughly the median" (*Amazon PPC bid range explained*, Amalyze advertising guide, https://amalyze.com/resources/guides/advertising/amazon-ppc-bid-range-explained , fetched 2026-11-09). Seller Metrics, through search-result summaries only (its pages returned no text to a plain GET), says suggested bids skew high and suggests starting 20–30% below them. [seat] Amazon's own pages say only "bids for ads similar to yours that have won impressions over the past 7 days"; the percentile reading is a vendor's inference, not Amazon's statement.
+
+### On bidding strategies
+
+- BidX (Lena Einschütz), *Case study on the dynamic bidding strategies of Amazon 2024*, 2024-04-17, https://www.bidx.io/blog/dynamic-bidding-strategies-amazon-2024 : about 130,000 campaigns in February 2024. Down-only campaigns showed better conversion rates and the lowest ACOS despite higher CPCs. BidX states the up-and-down ceiling as 100% for top of search and 50% elsewhere, and notes that a 900% top-of-search adjustment with up and down can in theory reach 20× the base bid (€1.50 → €30). Within the down-only campaigns, a first-page (top of search) adjustment above 0% went with the best CTR and conversion but about €0.30 higher CPC and "almost 5 percentage points" higher ACOS; a product-page adjustment above 0% went with the worst ACOS (21.65%). BidX recommends down only "if you are mainly aiming for profitability" and up and down for a product launch, and for manual keyword campaigns suggests halving the bid and allowing about +104% for first-page placements. The data are observational: advertisers chose their strategies, so the comparison carries selection effects (BidX's own caveat is weaker than this; the seat's reading).
+- Perpetua (Yasser Sakan), *Amazon Bidding: The Ultimate Guide to Dynamic Bidding Strategies and "Adjust Bids by Placement"*, 2019-02-27, https://perpetua.io/blog-amazon-ppc-advertising-dynamic-bidding-bid-strategy/ : applies the placement modifier first and the dynamic factor second (1 USD with +900% top of search under up and down gives 10 USD and a range up to 20 USD; product pages +400% give 5 USD and up to 7.50 USD under the 50% rule). It advises down only early, up and down for well-optimised campaigns, fixed for launches, and relays what it calls "Amazon’s word of caution": against comparing a new up-and-down campaign with an older down-only one, or copying a campaign to test a strategy ("When you copy a campaign, the sales history data is not transferred"), and against running down-only and up-and-down campaigns side by side to compare ("They will be bidding against each other in the same auction"). Dated 2019; the 50% figure is the older ceiling.
+- Novadata (Matthieu, CTO at Nova Analytics), *Amazon Tests ROAS Auto-Bidding for Sponsored Products*, 2026-02-18, https://novadata.io/resources/news/amazon-sponsored-products-roas-auto-bidding-2026 : reports a beta "Automatically optimizing ROAS within budget" option "rolling out to select US accounts": "You set a daily budget only. No base bid, no ROAS target." Its listed sources are "Hunter's Insights" and Amazon's Bid Controls API page; the latter (legacy v1 reference, read here) does not describe it. Unverified. A separate single-source item (Pulse, June 2026, read only as a search-result summary) reports a "Smart Bid V2" rollout to US accounts; no Amazon page read here mentions either.
+
+### On placements and top of search
+
+- Perpetua (Stephen Bench-Capon), *Maximizing the impact of top-of-search Amazon ad placements*, 2023-05-08, https://perpetua.io/blog-maximizing-the-impact-of-top-of-search-amazon-ad-placements/ : from Perpetua's Q1 2023 North American benchmark (for human readers only), top of search CTR 4.41%, conversion 13.8%, ACOS 27.0%; rest of search 0.79%, 8.7%, 31.8%; product pages 0.13%, 6.7%, 40.9%. Perpetua attributes the conversion gap partly to advertisers putting their best ASINs there and to shoppers being nearer a decision. It advises isolating keywords before applying large top-of-search multipliers and backing those campaigns with budget, and warns of cannibalization of organic sales for ASINs with a high organic rank: "If you see ACOS going down, but TACOS not following suit, this could mean your ads are negatively impacting your organic sales."
+  - [seat] The table compares placements across advertisers who chose to bid up where they convert best; it is not a causal effect of the placement. The research section (Narayanan & Kalyanam; Agarwal, Hosanagar & Smith) says naive position comparisons overstate the effect.
+- Perpetua help (read through search summaries): Perpetua computes a per-target share as top-of-search impressions for the keyword-ASIN pair over top-of-search impressions for the query, and targets with no clicks do not sync. Intentwise (blog, through search summaries) suggests roughly 10–15% top-of-search share as a rough figure for non-branded terms, "benchmarks vary by category and budget". Pacvue (a post titled "Three Reasons Winning Top of Search Doesn't Matter", through a search summary; the page returned 403) says a moderate base bid with a moderate multiplier tends to serve more often than a very low bid with a high multiplier.
+- Intentwise (Michael Waters), *New controls for offsite Sponsored Products placements*, 2025-04-16, https://www.intentwise.com/blog/new-controls-for-offsite-sponsored-products-placements : the controls are "Maximize reach" and "Minimize spend"; the placement report has an "Off Amazon" category with no breakdown by site.
+- Ad Badger, *Your Sponsored Products ads are now running inside creator reviews and nobody asked you first*, 2026-08-20, https://www.adbadger.com/blog/your-sponsored-products-ads-are-now-running-inside-creator-reviews-and-nobody-asked-you-first/ : creator (Influencer) placements began 2026-08-10 with campaigns enrolled by default; the console options are "Increase reach" (default) and "Limit off-Amazon spend"; top-of-search and product-page adjustments do not apply to creator placements; off-Amazon clicks share the campaign budget. Amazon's help page (F) confirms the last three points in its own words, under different option labels ("Increase reach off-Amazon (default)", "Limit reach to Amazon"). Ad Badger adds that "Only your base bid governs delivery on this inventory"; Amazon's page says instead that "Dynamic bidding and rule-based bidding still apply" off Amazon.
+
+### On dayparting
+
+- Perpetua (Stephen Bench-Capon), *Study: Intraday Optimization*, 2022-06-30, https://perpetua.io/blog-a-mini-study-intraday-optimization/ : US Sponsored Products, six weeks in spring 2022 (for human readers only). Average CPC fell from $1.13 to $0.84 over the day (Perpetua's hypothesis: competitors' budgets run out); conversion peaked around 20% between 6 and 10am Pacific against 6–8% overnight; ACOS sat at 20–23% from 6am to 6pm and reached 41% at 11pm. Perpetua recommends hourly bid multipliers.
+- Ad Badger (The Badger), *Amazon PPC Dayparting: The Complete Guide (2026)*, 2025-07-02 (modified 2026-08-12), https://www.adbadger.com/blog/how-to-use-dayparting-to-improve-your-amazon-ppc/ : answers "Does Amazon Ads support dayparting natively?" with "No ... Sellers use third-party tools like Ad Badger, or manual bid rules" (Amazon's schedule bid rules, H, exist but only raise bids), advises "Avoid turning ads off/on every hour", skipping dayparting during launches and high-traffic events, and lowering bids "instead of turning ads off entirely".
+  - [seat] The practitioner case for dayparting rests on hour-of-day differences in conversion and CPC. Amazon's down-only and up-and-down strategies already adjust each bid by predicted conversion, which includes the time of the auction as far as Amazon's model sees it; no source read here measures dayparting on top of dynamic bidding against dynamic bidding alone.
+
+## Published research
+
+Each item is [research]. Most of these studies are of Google, Yahoo, Bing, eBay or other marketplaces, not Amazon, and none uses Amazon's soft reserve as described in 2026. They describe mechanisms and measured effects; they do not set numbers for a skill. Where the seat read only an abstract or a search-result summary, the item says so.
+
+### Position effects in sponsored search
+
+- Agarwal, Hosanagar & Smith, "Location, Location, Location: An Analysis of Profitability of Position in Online Advertising Markets", *Journal of Marketing Research* 48(6), 2011, 1057–1073, doi:10.1509/jmr.08.0468 (bibliographic record checked through Crossref). From a field experiment with a retailer's search ads: click-through rate falls with position, but conversion rate does not fall the same way (it can rise lower down), and top positions cost more, so profit can be higher below the top. (Abstract and secondary summaries; the verbatim abstract was not retrieved.)
+- Narayanan & Kalyanam, "Position Effects in Search Advertising and their Moderators: A Regression Discontinuity Approach", *Marketing Science* 34(3), 2015, 388–407, doi 10.1287/mksc.2014.0893. Using the discontinuity at the rank boundaries of the auction: naive estimates overstate position effects; effects exist at some positions and not others; they are larger for smaller advertisers and for consumers with less prior experience of the keyword. (Abstract.)
+- Ghose & Yang, "An Empirical Analysis of Search Engine Advertising: Sponsored Search in Electronic Markets", *Management Science* 55(10), 2009, 1605–1622, doi 10.1287/mnsc.1090.1054. A six-month panel of a US retailer's Google keywords: conversion rates are highest at the top and fall with rank, so the value of a click differs by position; but "profits are often lower in the top positions than those in the middle positions due to the aggressive nature of bidding". Higher landing-page quality goes with higher conversion and lower CPC. (Abstract and NYU Stern press summary through search.)
+- Jerath, Ma, Park & Srinivasan, "A 'Position Paradox' in Sponsored Search Auctions", *Marketing Science* 30(4), 2011, 612–627, doi 10.1287/mksc.1110.0645. A theory with Korean search-engine data: a superior firm may bid below an inferior one, take a lower slot and still get more clicks, because consumers seek it out; the paradox appears in a large share of observed auctions. (Abstract through search.)
+- Ursu, "The Power of Rankings: Quantifying the Effect of Rankings on Online Consumer Search and Purchase Decisions", *Marketing Science* 37(4), 2018, 530–552, doi 10.1287/mksc.2017.1072. With randomised rankings at Expedia: rankings affect what consumers search, but conditional on search do not affect what they buy; the estimated position effect is smaller than non-experimental estimates. (Abstract and INFORMS release through search.)
+- Joo, Shi & Abhishek, "Do Sellers Benefit from Sponsored Product Listings? Evidence from an Online Marketplace", *Marketing Science* 43(4), 2024, 817–839, doi 10.1287/mksc.2021.0293. A large field study on a mobile marketplace: in the top slots shoppers favour organic over sponsored listings of the same product (a credibility effect), and are roughly indifferent lower down; a sponsored listing that displaces the seller's own top organic listing can hurt, while swapping a low organic listing for a top sponsored one can pay. (Abstract and university release through search.) The organic interaction is `areas/ads-organic-and-total.md`'s.
+- Blake, Nosko & Tadelis, "Consumer Heterogeneity and Paid Search Effectiveness: A Large-Scale Field Experiment", *Econometrica* 83(1), 2015, 155–174 (NBER Working Paper 20171, 2014), https://www.nber.org/papers/w20171 . eBay's experiments: "brand-keyword ads have no measurable short-term benefits"; for non-brand keywords "new and infrequent users are positively influenced by ads", while frequent buyers, who would have bought anyway, take most of the spend, so average returns were negative. Measured returns were a fraction of non-experimental estimates. (NBER abstract page.)
+- Rutz, Bucklin & Sonnier, "A Latent Instrumental Variables Approach to Modeling Keyword Conversion in Paid Search Advertising", *Journal of Marketing Research* 49(3), 2012, 306–319, doi 10.1509/jmr.10.0354 (abstract read through Crossref, 2026-11-09). Keyword-level paid-search data of a major lodging chain, with position treated as endogenous ("Position endogeneity in paid search data can arise from both omitted variables and measurement error") and corrected by a latent instrumental variable: "higher positions increase both the click-through and conversion rates. When advertisements are served in higher positions, approximately one-third of new conversions is due to increased click-through while approximately two-thirds are due to increased conversion rates."
+
+### Conversion rate by position: why the studies disagree
+
+The studies above give three different answers about conversion rate (not click-through rate, on which they agree) as an ad moves up. Abstracts re-read through Crossref on 2026-11-09:
+
+- Ghose & Yang (2009, a retailer's Google keywords, six months, hierarchical Bayesian simultaneous equations): "conversion rates are highest at the top and decrease with rank".
+- Rutz, Bucklin & Sonnier (2012, a lodging chain, latent instrumental variables for endogenous position): "higher positions increase both the click-through and conversion rates".
+- Agarwal, Hosanagar & Smith (2011, a field experiment on several keywords of an online retailer): "while click-through rate decreases with position, conversion rate increases with position and is even higher for more specific keywords"; in their usage a larger position number is lower on the page, so conversion rate rises lower down, and "the topmost position is not necessarily the revenue-or profit-maximizing position".
+- Narayanan & Kalyanam (2015, a regression discontinuity at the auction's rank boundaries, with competitors' data): "Because positions are determined through an auction, there are significant selection issues in measuring position effects. A simple mean comparison of outcomes at two positions is likely to be biased", and "experimentation is rendered difficult in this situation by competitors’ bidding behavior, which induces selection biases that cannot be eliminated by randomizing the bids for the focal advertiser"; their causal effects on click-through and sales orders are "much smaller, and exist only in some of the positions", stronger for smaller advertisers and less experienced consumers, weaker for brand- or product-specific keywords, exact-match copy and weekends.
+- Ursu (2018, randomised rankings at Expedia): position changes what is searched, but conditional on search it does not change what is bought.
+
+[seat] The disagreement is about method and data, not arithmetic. Two of the conversion findings (Ghose & Yang; Rutz et al.) come from observational panels with modelled endogeneity, one (Agarwal et al.) from a focal advertiser's own bid experiment, which Narayanan & Kalyanam argue cannot remove the selection that rivals' bids create; the one design built for that selection finds small position effects at some positions only, and the randomised-ranking study finds none on purchase given search. None is an Amazon study, all are text ads on web search or travel listings, and none has Amazon's top-of-search row of up to 5 (desktop) or 4 (mobile) product ads. What transfers is the warning: the conversion rate a seller's `campaignPlacement` report shows for top of search mixes the placement's effect with which shoppers and queries reach it and which rivals are beside it, and the direction of the true effect on conversion is not settled in the literature.
+
+### Position bias in click data and placement comparisons
+
+- Craswell, Zoeter, Taylor & Ramsey, "An experimental comparison of click position-bias models", WSDM 2008, 87–94, doi 10.1145/1341531.1341545 (Microsoft Research publication page read 2026-11-09). "A key source of bias is presentation order: the probability of click is influenced by a document’s position in the results page." They perturbed the ranking of a major search engine and tested four hypotheses: "The data are not well explained by simple position models, where some users click indiscriminately on rank 1 or there is a simple decay of attention over ranks. A ‘cascade’ model, where users view results from top to bottom and leave as soon as they see a worthwhile document, is our best explanation for position bias in early ranks." https://www.microsoft.com/en-us/research/publication/an-experimental-comparison-of-click-position-bias-models/
+- Joachims, Swaminathan & Schnabel, "Unbiased Learning-to-Rank with Biased Feedback", WSDM 2017, doi 10.1145/3018661.3018699 (arXiv 1608.04468, abstract read 2026-11-09; an IJCAI 2018 summary version is doi 10.24963/ijcai.2018/738). "position bias in search rankings strongly influences how many clicks a result receives, so that directly using click data as a training signal in Learning-to-Rank (LTR) methods yields sub-optimal results"; they give a counterfactual framework with a propensity-weighted ranking SVM "where click models take the role of the propensity estimator", tested on an operational search engine. https://arxiv.org/abs/1608.04468
+- Reiley, Li & Lewis, "Northern Exposure: A Field Experiment Measuring Externalities between Search Advertisements", ACM EC 2010, 297–304, doi 10.1145/1807342.1807390 (SSRN 2190189 abstract read through Crossref, 2026-11-09). A controlled experiment on the "north" ads above the organic results: "additional rival ads in the north tend to increase rather than decrease the click through rate (CTR) of the top sponsored listing". Neither the SSRN abstract nor the abstract of Lewis's MIT thesis (DSpace@MIT 1721.1/62403, read 2026-11-09), whose search chapter covers the same experiment, names the search engine; the brief that raised this gap calls it a Yahoo! experiment, which this seat did not confirm from a page read.
+- Xue, Liu, Du, Korlimarla & Wen, "Practical Lessons on Optimizing Sponsored Products in eCommerce", arXiv 2304.09107, 2023 (abstract page and PDF read 2026-11-09). The PDF's affiliation line is "Walmart eCommerce" (San Bruno, California; walmart.com addresses): this is Walmart's sponsored-products system, not Amazon's. "Based on our beforehand observation from our online shopping site, as the position increases, the CTR drops significantly. Such observation indicates that more clicks on an ad item is not necessarily equal to the preference of this item from users, it may be just because this ad item is in a good position." They de-bias by weighting training instances with functions of position ("polynomial, logarithmic, and inverse propensity weighting functions") and predict with position set aside. The follow-on PACC paper (Wang, Xue, Liu, Wen, Zhao, Guo & Yu, arXiv 2307.16060, SIGIR) by University of Illinois Chicago, Walmart eCommerce and Indeed authors de-biases click and conversion prediction jointly "on the E-commerce sponsored product search dataset".
+- Karra, Zhao, Murray & Pellegrini, "Nudging neural click prediction models to pay attention to position", CIKM 2023, listed on amazon.science (publication page read 2026-11-09): "users are more likely to interact with items that were shown prominently on a website. CTR models often overestimate the value of such items and show them more often, at the expense of items of higher quality that were previously shown at less prominent positions. This self-reinforcing position bias effect reduces both the immediate and long-term quality of recommendations"; they propose training that pushes networks to use position "(or other confounding variables) as much as possible to explain the training data" and a metric for bias. https://www.amazon.science/publications/nudging-neural-click-prediction-models-to-pay-attention-to-position
+  - [seat] The page names no Amazon product, so it shows Amazon researchers treating position as a confounder in click prediction, not how Amazon's sponsored-products ranker is built.
+- [seat] What this means for placement multipliers. The placement report compares a placement's CTR and CVR with another's, but each placement differs in where it sits on the page (position bias, cascade reading), in which shoppers and queries reach it (selection by Amazon's own relevance ranking and by the seller's bids), and in which rival ads are beside it (Reiley et al. find rivals can raise a top listing's CTR). The click-model and de-biasing literature corrects for position by modelling it or by randomising it; an advertiser's report does neither, and a seller cannot randomise Amazon's ranking. A multiplier judged on raw placement CTR or CVR rests on a biased comparison of unknown sign; the comparison that does identify a multiplier's effect is a change in the multiplier itself, read as `method/change-measurement.md` describes.
+
+### Auction design: second price, reserves and soft floors
+
+- Edelman, Ostrovsky & Schwarz, "Internet Advertising and the Generalized Second-Price Auction: Selling Billions of Dollars Worth of Keywords", *American Economic Review* 97(1), 2007, 242–259. The generalized second-price auction has no dominant strategy of truthful bidding; in its "locally envy-free" equilibria revenue is at least that of the VCG auction, with one such equilibrium matching VCG's positions and payments. (Abstract and summaries through search; drafts on the authors' sites.)
+- Ostrovsky & Schwarz, "Reserve Prices in Internet Advertising Auctions: A Field Experiment", *Journal of Political Economy* 131(12), 2023, 3352–3376, doi 10.1086/725702. On Yahoo, theory-based reserve prices raised revenue, most for keywords with high volume, high estimated reserves and few bidders. (Abstract through search; a secondary blog puts the average gain near 3%.)
+- Zeithammer, "Soft Floors in Auctions", *Management Science* 65(9), 2019, 4204–4221, doi 10.1287/mnsc.2018.3164. With symmetric bidders, soft floors in a second-price auction change neither allocation nor revenue (revenue equivalence); with asymmetric bidders revenue can go either way; profitable conditions are narrow, and a soft floor high enough effectively turns the auction into first price. (Abstract fragments and UCLA Anderson brief through search.)
+- Chen (Outreach Inc.), Nabi (Amazon Ads) & Siniscalchi (Amazon Ads and Northwestern University), "Advancing Ad Auction Realism: Practical Insights & Modeling Implications", AdKDD workshop at KDD 2023, arXiv 2307.11732; PDF at https://cdn.amazon.science/a5/3f/f4a65055483ba954a1bc476a17fc/advancing-ad-auction-realism-practical-insights-modeling-implications.pdf (full text read). They model advertisers as bandit learners in auctions where values vary by query, competitors are unobserved, feedback is partial, and "Pricing mechanisms are often incompletely specified. For instance, ad services may indicate that reserve prices are used, but not what they are, or how they are computed." Their soft floor: if the second bid is at or above the floor the price is the second bid; if the first is above the floor and the second below, the price is the floor; and "Crucially, if 𝑠 > 𝑏(1), then the high bidder still wins the auction ... and 𝑝 = 𝑏(1)". In a symmetric single-query setting soft floors had "virtually no impact on revenues"; in a multi-query setting where bidders choose which queries to target, formats gave different revenues, and "Soft-floor reserve prices can impact revenues; thus, our simulations provide some support to this common industry practice", with the ranking of first-price, second-price and soft-floor formats depending on the learning algorithm.
+  - [seat] The paper's soft-floor rule matches Amazon's 2026 statement in the case that matters most (a winner below the soft reserve still wins and pays its own bid). They differ in wording for a winner above the floor: the paper prices at the larger of the second bid and the floor, Amazon's statement says the winner "pay[s] the soft reserve". The paper is a simulation, two of whose three authors are at Amazon Ads, and says nothing about Amazon's production parameters.
+
+### Bid optimization and autobidding
+
+- Aggarwal, Badanidiyuru & Mehta, "Autobidding with Constraints", WINE 2019, LNCS 11920, 17–30. For an advertiser maximising value under spend or return constraints, uniform bidding (each bid a constant multiple of value) is optimal when the auction is truthful, and not in general otherwise. (Through search and later papers' summaries.)
+- Balseiro, Deng, Mao, Mirrokni & Zuo, "Robust Auction Design in the Auto-bidding World", arXiv 2111.02468, 2021 (abstract read on arXiv). With value-maximizing bidders under return-on-spend constraints, "by choosing reserve prices appropriately the auctioneer can improve not only the total revenue but also the total welfare"; the results generalize to VCG, GSP and first-price auctions, and additive boosts improve welfare further; an empirical study on online advertising auction data accompanies the theory.
+  - [seat] Amazon's rule-based bidding (a ROAS guardrail, bids up to 5× the adjusted bid) is a value-maximiser under a return-on-spend constraint in exactly this sense, and its up-and-down strategy scales bids by predicted conversion, which is a form of value-proportional bidding. The papers say such bidding is optimal in truthful auctions; a GSP with soft reserves is not truthful.
+- Jeunen, Murphy & Allison (Amazon), "Off-Policy Learning-to-Bid with AuctionGym", KDD 2023, doi 10.1145/3580305.3599877 (an earlier version, "Learning to Bid with AuctionGym", won the AdKDD 2022 best paper). Bid data logged from one's own past policy carry unobserved confounding, so naive learning from them is biased; the paper proposes off-policy methods and a simulator. The simulator, https://github.com/amzn/auction-gym (README read), simulates second-price auctions and first-price auctions with bid shading. The repository makes no claim about Amazon's production auction. (PDF at https://cdn.amazon.science/4f/3c/7f9a5c6c4181894d5e64c684c0d7/learning-to-bid-with-auctiongym.pdf was not extracted.)
+- Jeunen, Stavrogiannis, Sayedi & Allison (Amazon), "A probabilistic framework to learn auction mechanisms via gradient descent", AAAI 2023 Workshop on AI for Web Advertising, https://www.amazon.science/publications/a-probabilistic-framework-to-learn-auction-mechanisms-via-gradient-descent (publication page read). Learns auction rules by gradient descent over a probabilistic relaxation of allocation and pricing.
+  - [seat] The confounding point in AuctionGym is the research form of a practical warning: a seller's own history of "bid X gave Y" was produced by the seller's earlier bidding choices, so it does not by itself identify what a different bid would do. `method/change-measurement.md` owns how cx reads a change.
+
+## Stable enough for a skill
+
+These are mechanisms Amazon states on more than one current page, or states once in a specification that a write would hit. Figures inside them are still read live.
+
+1. A bid is a maximum. The price of a click (or of 1,000 viewable impressions) never exceeds the maximum effective bid, which is the base bid after adjustments and strategy. Exception: Sponsored Brands and display cost control set a target average, not a maximum. Sources: *Understand Amazon Ads auctions*; *Understand bidding on Sponsored ads*; *What is cost per click (CPC)?*; *Understand Cost controls for sponsored ads*.
+2. Amazon's auction ranks by relevance and bid and prices as a generalized second-price auction with a hard reserve (entry threshold) and soft reserves (which can raise the price above the runner-up, up to the winner's maximum). Sources: *Understand Amazon Ads auctions*; *Amazon's response to the FTC's lawsuit regarding Sponsored Ads*; *Understand auctions for Sponsored Products ads across retailers*.
+3. A higher bid does not guarantee a win: lower bids with higher relevance win. Sources: *Understand Amazon Ads auctions*; *Amazon's response to the FTC's lawsuit*.
+4. Sponsored Products has four strategies: down only (`LEGACY_FOR_SALES`, the API default), up and down (`AUTO_FOR_SALES`), fixed (`MANUAL`) and rule-based (`RULE_BASED`). Sources: *Bidding strategies for Sponsored Products*; Sponsored Products 3.0 OpenAPI.
+5. Down only can lower a bid by up to 100%; fixed never moves it; up and down can raise it by up to 100% (current help, guide and model say for all placements). Sources: *Bidding strategies for Sponsored Products*; *Dynamic bidding for Sponsored Products*; Sponsored Products 3.0 OpenAPI.
+6. Sponsored Products placement adjustments are increases from 0% to 900% for top of search, rest of search, product pages and Amazon Business, and dynamic bidding then applies to the placement-adjusted bid. Sources: *Understand bid adjustments in Sponsored ads*; *Adjust Sponsored Products bids*; Sponsored Products 3.0 OpenAPI; *Sponsored Products best practices*.
+7. Placement, audience and video adjustments multiply with each other; among several audiences only the highest boost applies. Sources: *Adjust Sponsored Products bids*; *Understand bid adjustments in Sponsored ads*; Sponsored Products 3.0 OpenAPI (`shopperCohortBidding`).
+8. Off-Amazon placements use the same bid and budget; top-of-search, rest-of-search and product-page adjustments do not apply there; dynamic and rule-based bidding do. Source: *Understand Sponsored Products off-Amazon advertising*.
+9. Rule-based bidding moves base bids under a ROAS guardrail, up to 5× the adjusted bid, with no guarantee; it excludes schedule bid rules on the same campaign. Sources: *Bidding rules for Sponsored Products*; *Sponsored Products Rule-based bidding overview*.
+10. Schedule bid rules only raise bids (1–100% of base), add to each other, add to the placement adjustment, take effect within 24 hours, and are limited to 20 per campaign. Sources: *Bidding rules for Sponsored Products*; *Scheduled bid rules* (API guide).
+11. Suggested bids and ranges are computed from recent winning bids of similar ads (7 days on the help page), updated daily, and are not forecasts. Sources: *Understand bidding on Sponsored ads*; *Theme-based bid suggestions - Quick-start guide*; *Performance Estimate (beta)*; Sponsored Display 3.0 OpenAPI.
+12. Sponsored Products has no separate placement report in the v3 API; placement is a grouping of the campaign report. Top-of-search impression share is campaign-level in v1 and also target-level in v3. Sources: *Placement reports*; *Reporting columns (version 3)*; v1 *Search (Impression share)* page.
+13. Sponsored Brands adjusts placements only for top of search and rest of search (other than top of search); automated bidding optimizes placements below top of search; negative placement adjustments are no longer accepted from 2026-06-15. Sources: *Understand bid adjustments in Sponsored ads*; *Create a Sponsored Brands campaign*; Sponsored Brands 4.0 OpenAPI; Ads API release notes.
+14. Sponsored Brands "drive page visits" and display ads can raise a bid by up to 300% for an auction. Sources: *Understand bidding on Sponsored ads*; *Optimization strategies for Display*; *Amazon Ads API for Sponsored Display overview*.
+15. A display optimization strategy, a Sponsored Brands goal and a site restriction cannot be changed after creation. Sources: *Optimization strategies for Display*; *Campaign goals for Sponsored Brands*; *Sites selection for Sponsored Ads campaigns*; Sponsored Products 3.0 and Sponsored Brands 4.0 OpenAPI.
+16. Click data arrive within about 12 hours, clicks can be invalidated for 72 hours, and Sponsored Products conversions can be restated for 60 days. Sources: *Limits, constraints, and quotas*; *Sponsored Products performance datasets*.
+17. Bid writes are batch writes answered item by item: Sponsored Products `PUT /sp/keywords`, `/sp/targets`, `/sp/adGroups`, `/sp/campaigns` (versioned vendor media types); v1 `POST /adsApi/v1/update/{targets,adGroups,campaigns}` (`application/json`); Sponsored Brands `PUT /sb/keywords`, `/sb/targets`, `/sb/v4/campaigns`; Sponsored Display `PUT /sd/targets`, `/sd/adGroups`. A refused item comes back with its request index and a typed reason, and for out-of-range bids the limits. Sources: Sponsored Products 3.0, v1 merged and per-product, Sponsored Brands 3.0 and 4.0, Sponsored Display 3.0 OpenAPI.
+18. Under an enabled optimization rule (Sponsored Brands campaign, display ad group), a target's or keyword's bid is not the bid the auction uses. Sources: Sponsored Brands 3.0 OpenAPI; Sponsored Display 3.0 OpenAPI; *Understand Cost controls for sponsored ads*.
+19. Existing sponsored ads campaigns serve in Alexa for Shopping (Rufus and Alexa+ merged, May 2026), as prompts, and in Alexa+ Conversational Ads without set-up; Amazon publishes no rule on which bid adjustments apply there. Sources: *What agentic shopping means for advertising*; *Meet Alexa for Shopping*; *Understand prompts on Sponsored Products and Sponsored Brands*.
+
+## What cx must read at run time
+
+All reads go through the seller's own connection (the `cx raw amazon-ads` door, `method/ads-api-surface.md`); this seat called none of these hosts.
+
+- The campaign's strategy and adjustments: Sponsored Products `dynamicBidding.strategy`, `dynamicBidding.placementBidding[]`, `dynamicBidding.shopperCohortBidding[]`, `offAmazonSettings`, `siteRestrictions` (`POST /sp/campaigns/list`); Sponsored Brands `bidding.bidOptimization`, `bidAdjustmentsByPlacement`, `shopperCohortBidAdjustments`, `goal`, `costType` (`POST /sb/v4/campaigns/list`); display `bidOptimization`, `costType` (`/sd/campaigns`, `/sd/adGroups`); or the v1 `bidSettings` on `/adsApi/v1/...` campaigns where the account has moved to v1.
+- The bids themselves: ad group `defaultBid`, keyword and target `bid` (Sponsored Products `/sp/keywords/list` and `/sp/targets/list`, Sponsored Brands keywords and targets, display targets).
+- The marketplace's live accepted range: a write's error (`BID_OUT_OF_MARKET_PLACE_RANGE`, `BID_GT_BUDGET`, `BID_INVALID_PLACEMENT`) is the authority; the limits page is a human reference only.
+- Rules on the campaign: `POST /sp/rules/optimization/search` (schedule and performance rules), `GET /sp/rules/campaignOptimization/{id}` and `POST /sp/rules/campaignOptimization/state` (rule-based bidding), `POST /sp/rules/campaignOptimization/eligibility` (whether rule-based bidding can be applied now), `/sb/rules/optimization/list`, `/sd/adGroups/{adGroupId}/optimizationRules`.
+- Bid recommendations for the targets at hand: `POST /sp/targets/bid/recommendations` (themes, ranges, and with the v5 content type estimated impressions and the `includeAnalysis` bid points where offered), `/sd/targets/bid/recommendations`, `/sb/recommendations/optimization`, the campaign recommendations `/sp/campaign/recommendations`, and the v1 recommendations list or query where available. A 422 from the Sponsored Products endpoint means the marketplace or target is not supported.
+- Placement performance: `spCampaigns` grouped by `campaignPlacement` (with `campaignSite` for Amazon Business), `sbCampaignPlacement`, or v1 reports with `placementClassification.value`; top-of-search impression share from `spCampaigns` (grouped by `campaign` or `campaignPlacement`), `spTargeting`, `sbCampaigns`, `sbTargeting` (v3) or campaign-level v1. Prompt clicks and their cost: `spPromptAdExtension` and `sbPromptAdExtension` grouped by `promptAdExtension` (US).
+- Hour-of-day performance, where a schedule is in question: v1 `hour.value` (14 days), Stream v2 `ads-performance-v1` (`hour.value`), or the Stream v1 datasets `sp-traffic` and `sp-conversion` (the ids a subscription takes; both stop delivering on 2027-06-30) if the seller has a subscription.
+- The bid writes, as the models give them (O): Sponsored Products `PUT /sp/keywords` (`keywords[].keywordId`, `.bid`; `application/vnd.spKeyword.v3+json`), `PUT /sp/targets` (`targetingClauses[].targetId`, `.bid`; `application/vnd.spTargetingClause.v3+json`), `PUT /sp/adGroups` (`adGroups[].adGroupId`, `.defaultBid`; `application/vnd.spAdGroup.v3+json`), `PUT /sp/campaigns` (`campaigns[].dynamicBidding.strategy`, `.placementBidding[]`, `.shopperCohortBidding[]`; `application/vnd.spCampaign.v3+json`); v1 `POST /adsApi/v1/update/targets` (`targets[].bid.bid`), `update/adGroups` (`adGroups[].bid.defaultBid`), `update/campaigns` (`campaigns[].optimizations.bidSettings.bidStrategy` and `.bidAdjustments`), all `application/json`; Sponsored Brands `PUT /sb/keywords` (array of `{keywordId, adGroupId, campaignId, bid}`), `PUT /sb/targets` (`targets[]`), `PUT /sb/v4/campaigns` (`campaigns[].bidding.bidOptimization`, `.bidAdjustmentsByPlacement[]`, `.shopperCohortBidAdjustments[]`; `application/vnd.sbcampaignresource.v4+json`); Sponsored Display `PUT /sd/targets` and `PUT /sd/adGroups` (`defaultBid`). The batch ceiling is the model's `maxItems` (or the operation text, where the SD model gives two figures); the outcome of each item is the 207 body's `success`/`error` entry for its request index (`targetRequestIndex` for SB targets, request order for SB keywords and SD), and the reason codes and `lowerLimit`/`upperLimit` in a bidding error are the live range.
+- After a write: a read-back of the entity (`POST /sp/keywords/list`, `/sp/targets/list`, `/adsApi/v1/query/targets`, `GET /sd/targets` and so on), and, for display ad groups and Sponsored Brands campaigns, the attached optimization rules (`/sd/adGroups/{adGroupId}/optimizationRules`, `/sb/rules/optimization/list`), because a bid under an enabled rule is not the auction's bid.
+- The marketplace and the account's time zone (profile), because v3 reports and schedule windows run in it; v1 reports aggregate by each campaign's time zone (`timeZone.value`).
+- Whether an account is in a marketplace where a feature exists (off-Amazon list, Amazon Business list, audience count outside the US, video adjustment US-only): read from the account's own responses, not from these lists.
+
+## Pages read
+
+All read 2026-11-09. "Search" in the type column means the item was read only through a search engine's result summary, not fetched.
+
+| Page | URL | Type | Date read | What it gave |
+|---|---|---|---|---|
+| Understand Amazon Ads auctions (help GEXGZNBBWPZZK7P8) | https://advertising.amazon.com/help/GEXGZNBBWPZZK7P8 | official help | 2026-11-09 | Relevance factors; hard and soft reserves; price may exceed runner-up, never the maximum |
+| Understand bidding on Sponsored ads (help GDS2E64R9GQ35768) | https://advertising.amazon.com/help/GDS2E64R9GQ35768 | official help | 2026-11-09 | Bid, effective bid, maximum effective bid; CPC and vCPM by product; SB 300%; vCPM minimums; suggested bid from 7 days of winning bids |
+| Understand auctions for Sponsored Products ads across retailers (help GW3CP75ZUTET4QEF) | https://advertising.amazon.com/help/GW3CP75ZUTET4QEF | official help | 2026-11-09 | Retailer reserves; price adjusted to retailer demand within the maximum |
+| Bidding strategies for Sponsored Products (help GCU2BUWJH2W3A8Z7) | https://advertising.amazon.com/help/GCU2BUWJH2W3A8Z7 | official help | 2026-11-09 | Four strategies; 100% for all placements; audiences up to 10 |
+| Understand bid adjustments in Sponsored ads (help GQ4R4FR4H56JL6V8) | https://advertising.amazon.com/help/GQ4R4FR4H56JL6V8 | official help | 2026-11-09 | Placement definitions; stacking examples; SB only top and rest of search |
+| Set bids in a Sponsored Products campaign (help GTMXQWASBRZTHLF2) | https://advertising.amazon.com/help/GTMXQWASBRZTHLF2 | official help | 2026-11-09 | Keyword bid precedence; 45-day event suggestions; video 0–900% |
+| Adjust Sponsored Products bids (help GYYZVM7LGSRYGWV5) | https://advertising.amazon.com/help/GYYZVM7LGSRYGWV5 | official help | 2026-11-09 | Highest audience wins; browse placements across retailers; video US only |
+| Bidding rules for Sponsored Products (help GDLQ5D2BNFCAU3TW) | https://advertising.amazon.com/help/GDLQ5D2BNFCAU3TW | official help | 2026-11-09 | Rule-based up to 5×; schedule rules additive; hourly report 30/14 days |
+| Apply Schedule bid rules for Sponsored Products (help G7JFAU9C7PHA3Q68) | https://advertising.amazon.com/help/G7JFAU9C7PHA3Q68 | official help | 2026-11-09 | Strategies that accept schedule rules; event schedules |
+| Sponsored Brands bids by placement and audiences (help GC8K522SMNCPH99Z) | https://advertising.amazon.com/help/GC8K522SMNCPH99Z | official help | 2026-11-09 | SB decrease example; one audience; cost control is an average |
+| Understand Cost controls for sponsored ads (help GNJDAT6AEYD3XJA8) | https://advertising.amazon.com/help/GNJDAT6AEYD3XJA8 | official help | 2026-11-09 | Target average CPC; SB 7-day rolling average; Display: no manual bids, no suggested bids |
+| Optimization strategies for Display (help GDRFG6U7TRD72EFW) | https://advertising.amazon.com/help/GDRFG6U7TRD72EFW | official help | 2026-11-09 | 300%; reach/page visits/conversions/leads; vCPM below $3 warning; no change after creation |
+| Performance Estimate (beta) (help GHFYQ2D4LQC5ABPW) | https://advertising.amazon.com/help/GHFYQ2D4LQC5ABPW | official help | 2026-11-09 | Impression estimates by bid; not a forecast; budget ignored |
+| Understand Sponsored Products off-Amazon advertising (help GYTD2Z3SYMAAMVXA) | https://advertising.amazon.com/help/GYTD2Z3SYMAAMVXA | official help | 2026-11-09 | Off-Amazon countries, routes, CPC setting; placement adjustments not applied; reporting labels |
+| Sites selection for Sponsored Ads campaigns (help GJGYD2FKNF3MUWBQ) | https://advertising.amazon.com/help/GJGYD2FKNF3MUWBQ | official help | 2026-11-09 | Amazon Business and Haul sites; no cohort or AB adjustments there; not changeable |
+| Understand prompts on Sponsored Products and Sponsored Brands (help GLWNW26EWQZW46WZ) | https://advertising.amazon.com/help/GLWNW26EWQZW46WZ | official help | 2026-11-09 | Prompts CPC, auto-enrolled in US |
+| Campaign goals for Sponsored Brands (help GXXPFXLSVX9VD3U9) | https://advertising.amazon.com/help/GXXPFXLSVX9VD3U9 | official help | 2026-11-09 | Goals fixed; TOS IS per reserved keyword; reserve SOV terms |
+| Create a Sponsored Brands campaign (help GF86HBCNDJUAC5WN) | https://advertising.amazon.com/help/GF86HBCNDJUAC5WN | official help | 2026-11-09 | Automated bidding below top of search; increase or decrease |
+| Downloadable reports in sponsored ads (help GBYSPTSLR337JMLH) | https://advertising.amazon.com/help/GBYSPTSLR337JMLH | official help | 2026-11-09 | Campaign placement, keyword placement, search term impression share reports |
+| Bidding and budget (help G8MCJMVDGD6VQSPW), Support Center homepage (GX4827LQGBFHA2U8) and category nodes | https://advertising.amazon.com/help/G8MCJMVDGD6VQSPW | official help index | 2026-11-09 | Index of help article ids used above |
+| Amazon's response to the FTC's lawsuit regarding Sponsored Ads | https://www.aboutamazon.com/md/company-news/amazon-ftc-sponsored-ads-lawsuit-response | official corporate | 2026-11-09 | GSP; hard and soft reserves; ranking; 92% and 128th-bid figures |
+| What is cost per click (CPC)? | https://advertising.amazon.com/library/guides/cost-per-click | official library | 2026-11-09 | Final CPC from adjusted bid plus factors; maximum CPC |
+| What is cost per mille (CPM)? | https://advertising.amazon.com/library/guides/cost-per-mille | official library | 2026-11-09 | CPM and vCPM background |
+| Sponsored Products best practices | https://advertising.amazon.com/library/guides/sponsored-products-best-practices | official library | 2026-11-09 | Adjusted bid is the maximum; placements to 900% |
+| An author's guide to Sponsored Products | https://advertising.amazon.com/library/guides/authors-guide-to-sponsored-products | official library | 2026-11-09 | 100%/50% up-and-down wording; bid-only example |
+| Dynamic bidding for Sponsored Products | https://advertising.amazon.com/library/guides/dynamic-bidding-sponsored-products | official library | 2026-11-09 | 100% all placements; spend-double warning; two-week review |
+| Targeting with Sponsored Products | https://advertising.amazon.com/library/guides/targeting-with-sponsored-products | official library | 2026-11-09 | Match-type bid ordering; strategy by campaign type |
+| Sponsored Brands for authors | https://advertising.amazon.com/library/guides/sponsored-brand-authors | official library | 2026-11-09 | SB background for KDP |
+| Sponsored Products (product page) | https://advertising.amazon.com/solutions/products/sponsored-products | official product page | 2026-11-09 | Bid is a maximum; placements beyond Amazon; budgets not paced |
+| Sponsored Brands (product page) | https://advertising.amazon.com/solutions/products/sponsored-brands | official product page | 2026-11-09 | CPC, vCPM, reserve SOV with minimum commitment |
+| Sponsored Display (product page) | https://advertising.amazon.com/solutions/products/sponsored-display | official product page | 2026-11-09 | Renamed display ads; Ads Agent |
+| Improve your campaign performance on rest of search placements | https://advertising.amazon.com/resources/whats-new/improve-campaign-performance | official what's new | 2026-11-09 | Rest-of-search adjustment |
+| Helping customers discover new selection with Sponsored Products on premium apps and websites | https://advertising.amazon.com/resources/whats-new/sponsored-products-appear-on-publishers | official what's new | 2026-11-09 | First publishers; placement reporting |
+| Search term impression share report for Sponsored Products | https://advertising.amazon.com/resources/whats-new/search-term-impression-report-sponsored-products | official what's new | 2026-11-09 | Search term impression share and rank |
+| Top-of-search impression share metric | https://advertising.amazon.com/resources/whats-new/top-of-search-impression-share-metric | official what's new | 2026-11-09 | Definition; 90-day look-back; adjust placements to raise it |
+| Sponsored Display launches bid recommendations | https://advertising.amazon.com/resources/whats-new/sponsored-display-bid-recommendations | official what's new | 2026-11-09 | Winning bids of similar ads; daily updates |
+| Sponsored Products 3.0 OpenAPI | https://d1y2lf8k3vrkfu.cloudfront.net/openapi/en-us/dest/SponsoredProducts_prod_3p.json | official spec | 2026-11-09 | Strategies, placements 0–900, cohorts, off-Amazon, site restrictions, recommendations, errors |
+| Amazon Ads API v1 merged OpenAPI (and the prod_3p_ALL variant) | https://d1y2lf8k3vrkfu.cloudfront.net/openapi/en-us/dest/AmazonAdsAPIALLMerged_prod_3p.json | official spec | 2026-11-09 | BidStrategy, Placement, BidSettings, CostType, KPI, BID recommendations |
+| Sponsored Brands 4.0 OpenAPI | https://d3a0d0y2hgofx6.cloudfront.net/openapi/en-us/sponsored-brands/4-0/openapi.json | official spec | 2026-11-09 | bidOptimization, −99..900 placements, goals, cost types |
+| Sponsored Brands 3.0 OpenAPI | https://d3a0d0y2hgofx6.cloudfront.net/openapi/en-us/sponsored-brands/3-0/openapi.yaml | official spec | 2026-11-09 | Legacy bidMultiplier −99..99 |
+| Sponsored Display 3.0 OpenAPI | https://d3a0d0y2hgofx6.cloudfront.net/openapi/en-us/sponsored-display/3-0/openapi.yaml | official spec | 2026-11-09 | bidOptimization, optimization rules, bid recommendations |
+| Recommendations OpenAPI | https://dtrnk0o2zy01c.cloudfront.net/openapi/en-us/dest/Recommendations_prod_3p.json | official spec | 2026-11-09 | Bid-related recommendation types |
+| Amazon Ads MCP campaign management spec (static) | https://d3a0d0y2hgofx6.cloudfront.net/en-us/mcp/campaign_management.json | official spec | 2026-11-09 | Tool shapes for bid settings (static file, not the server) |
+| Sponsored Products Rule-based bidding overview | https://advertising.amazon.com/API/docs/en-us/guides/rules/bidding-rules/overview | official API guide | 2026-11-09 | Eligibility, guardrail, auto-disable, minimum budgets |
+| Getting started with the Sponsored Products rule-based bidding API | https://advertising.amazon.com/API/docs/en-us/guides/rules/bidding-rules/getting-started | official API guide | 2026-11-09 | Endpoints, recurrence, ADOPT action, eligibility (content type, refresh and ROAS-only wording are in the SP 3.0 OpenAPI) |
+| Scheduled bid rules | https://advertising.amazon.com/API/docs/en-us/guides/rules/bidding-rules/scheduled-bid-rules | official API guide | 2026-11-09 | Schedule rule model, stacking, 20 per campaign, 24 hours (1–100 value and PERFORMANCE subcategory are in the SP 3.0 OpenAPI) |
+| Schedule-based budget rules | https://advertising.amazon.com/API/docs/en-us/guides/rules/budget-rules/schedule-based | official API guide | 2026-11-09 | Budget counterpart (left to budgets-and-pacing) |
+| Theme-based bid suggestions - Quick-start guide | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-products/bid-suggestions/theme-based-bid-suggestions-quickstart-guide | official API guide | 2026-11-09 | Computation; impact metrics caveats |
+| Creating manual campaigns (Sponsored Products) | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-products/get-started/manual-campaigns | official API guide | 2026-11-09 | Where bid recommendations sit in campaign creation |
+| Sponsored Products features | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-products/features | official API guide | 2026-11-09 | Feature list |
+| Keyword group targeting | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-products/keyword-group-targeting | official API guide | 2026-11-09 | Keyword groups in v4 recommendations |
+| Limits, constraints, and quotas | https://advertising.amazon.com/API/docs/en-us/reference/concepts/limits | official API reference | 2026-11-09 | Min/max bids; 12h and 72h |
+| Bid controls (legacy reference) | https://advertising.amazon.com/API/docs/en-us/reference/1/bid-controls | official API reference | 2026-11-09 | Bid+ history |
+| Ads API release notes | https://advertising.amazon.com/API/docs/en-us/release-notes/index | official API notes | 2026-11-09 | Dated changes cited throughout |
+| Deprecations (added by the verifier) | https://advertising.amazon.com/API/docs/en-us/release-notes/deprecations | official API notes | 2026-11-09 | v3 reporting, Stream v1 and SB v2 reporting end 2027-06-30; v2 bid recommendations and v3 theme content type shut off 2025-05-15 |
+| Campaign reports (v3) | https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/report-types/campaign | official API guide | 2026-11-09 | campaignPlacement grouping, retention |
+| Targeting reports (v3) | https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/report-types/targeting | official API guide | 2026-11-09 | topOfSearchImpressionShare, keywordBid |
+| Placement reports (v3) | https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/report-types/placement | official API guide | 2026-11-09 | No separate SP placement report; sbCampaignPlacement |
+| Bid adjustments reports (v3) | https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/report-types/bid-adjustment | official API guide | 2026-11-09 | DSP only |
+| Audience reports (v3) | https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/report-types/audience | official API guide | 2026-11-09 | Audience reporting |
+| Reporting columns (version 3) | https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/columns | official API guide | 2026-11-09 | topOfSearchImpressionShare definition and report types |
+| v1 placement and placement classification dimensions | https://advertising.amazon.com/API/docs/en-us/guides/reporting/ads-v1/dimensions/supply/placement-classification | official API guide | 2026-11-09 | placementClassification.value |
+| Search (Impression share), v1 metrics | https://advertising.amazon.com/API/docs/en-us/guides/reporting/ads-v1/metrics/impression-share/search | official API guide | 2026-11-09 | TOS IS compatibility |
+| v1 hour dimension | (ads-v1 dimensions, time) | official API guide | 2026-11-09 | hour.value products |
+| v1 costs metrics | https://advertising.amazon.com/API/docs/en-us/guides/reporting/ads-v1/metrics/costs-and-fees/costs | official API guide | 2026-11-09 | Cost metric names |
+| Dates and historical data retention (v1) | https://advertising.amazon.com/API/docs/en-us/guides/reporting/ads-v1/dates | official API guide | 2026-11-09 | Hour 14 days; time zone |
+| The campaign entity in the Amazon Ads API (campaign management) | https://advertising.amazon.com/API/docs/en-us/guides/campaign-management/entities/campaign | official API guide | 2026-11-09 | bidSettings by product; SD later |
+| Campaign management example payloads | (campaign-management guides) | official API guide | 2026-11-09 | Bid settings shapes |
+| Sponsored Products & Brands: Audience Bid Boosting (usage example) | https://advertising.amazon.com/API/docs/en-us/knowledge-hub/blogs/usage-examples/2026-01-sp-sb-bid-boosting | official API blog | 2026-11-09 | spAudiences and sbAudiences grouped by campaign_bid_boost_segment |
+| Shopper cohort bidding (Sponsored Brands) | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-brands/bidding/shopper-cohort | official API guide | 2026-11-09 | Cohort with cost control and placements |
+| Getting started with campaigns (Sponsored Brands) | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-brands/campaigns/get-started-with-campaigns | official API guide | 2026-11-09 | Automated bidding below TOS; tip on smart defaults |
+| Sponsored Brands features | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-brands/features | official API guide | 2026-11-09 | Feature list |
+| Managing Sponsored Brands optimization rules | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-brands/campaigns/managing-optimization-rules | official API guide | 2026-11-09 | CPC rule for PAGE_VISIT; targets may be added; 21-day advice |
+| Reserve share of voice (Sponsored Brands) | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-brands/reserve-share-of-voice | official API guide | 2026-11-09 | Fixed price; deal states |
+| Amazon Ads API for Sponsored Display overview | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-display/overview | official API guide | 2026-11-09 | 300%; strategy table; vCPM floors |
+| Sponsored Display features | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-display/features | official API guide | 2026-11-09 | Feature list |
+| Sponsored Products budget recommendations and missed opportunities | https://advertising.amazon.com/API/docs/en-us/guides/sponsored-products/budget-recommendations-and-missed-opportunities | official API guide | 2026-11-09 | Budget side (left to budgets-and-pacing) |
+| Partner opportunities catalog: campaign optimization | https://advertising.amazon.com/API/docs/en-us/guides/recommendations/partner-opportunities/catalog/campaign-optimization | official API guide | 2026-11-09 | Recommendation catalogue context |
+| Sponsored Products performance datasets (Marketing Stream v1) | https://advertising.amazon.com/API/docs/en-us/guides/amazon-marketing-stream/datasets/sp-performance | official API guide | 2026-11-09 | Hourly placement data; 12h, 72h, 60 days |
+| Overview (Amazon Marketing Stream Version 2) | https://advertising.amazon.com/API/docs/en-us/guides/amazon-marketing-stream/v2/overview | official API guide | 2026-11-09 | 15-minute delivery; totals, not deltas |
+| Ads performance (Stream v2 dataset) | https://advertising.amazon.com/API/docs/en-us/guides/amazon-marketing-stream/v2/datasets/ads-performance | official API guide | 2026-11-09 | Hourly grain; 70-day restatement window; placement classification; no impression share |
+| AI Migration Skills | https://advertising.amazon.com/API/docs/en-us/mcp/ai-migration-skills | official API guide | 2026-11-09 | ads-api-migration skill |
+| Download Skills from the Amazon Ads MCP Server | https://advertising.amazon.com/API/docs/en-us/mcp/skills/get-started | official API guide | 2026-11-09 | Skills download path (on an Ads API host) |
+| FTC, States Sue Amazon Over Secret Ad Surcharge Scheme | https://www.ftc.gov/news-events/news/press-releases/2026/08/ftc-states-sue-amazon-over-secret-ad-surcharge-scheme | regulator (plain GET returned a "Page Not Found" shell; read through a fetch tool by the verifier) | 2026-11-09 | Soft-reserve allegation; 30–40%, 70%, 80% |
+| BidX, Case study on the dynamic bidding strategies of Amazon 2024 | https://www.bidx.io/blog/dynamic-bidding-strategies-amazon-2024 | practitioner | 2026-11-09 | 130,000 campaigns; strategy and placement comparisons |
+| Trellis, Amazon Placement Bid Adjustments | https://gotrellis.com/resources/blog/amazon-ad-placements | practitioner | 2026-11-09 | Max CPC per placement formula; ceiling, not target |
+| Scale Insights, Amazon Bid Modifiers | https://scaleinsights.com/learn/bid-modifiers-top-of-search-rest-of-search-and-product-placements | practitioner | 2026-11-09 | Qualitative placement guidance (states no up-and-down ceiling) |
+| Perpetua, Study: Intraday Optimization | https://perpetua.io/blog-a-mini-study-intraday-optimization/ | practitioner | 2026-11-09 | Hour-of-day CPC, CVR, ACOS |
+| Perpetua, Amazon Bidding: The Ultimate Guide to Dynamic Bidding Strategies | https://perpetua.io/blog-amazon-ppc-advertising-dynamic-bidding-bid-strategy/ | practitioner | 2026-11-09 | Order of placement then dynamic; strategy advice (2019) |
+| Perpetua, Maximizing the impact of top-of-search Amazon ad placements | https://perpetua.io/blog-maximizing-the-impact-of-top-of-search-amazon-ad-placements/ | practitioner | 2026-11-09 | Placement benchmark table; cannibalization warning |
+| Ad Badger, Amazon PPC Dayparting: The Complete Guide (2026) | https://www.adbadger.com/blog/how-to-use-dayparting-to-improve-your-amazon-ppc/ | practitioner | 2026-11-09 | Dayparting views |
+| Ad Badger, creator placements | https://www.adbadger.com/blog/your-sponsored-products-ads-are-now-running-inside-creator-reviews-and-nobody-asked-you-first/ | practitioner | 2026-11-09 | Creator placements from 2026-08-10 |
+| Ad Badger, Calculating a Bid Strategy for Amazon Ad Campaigns | https://www.adbadger.com/blog/calculating-amazon-bid-strategy/ | practitioner | 2026-11-09 | RPC × target ACOS |
+| Ad Badger, Amazon Says Its Ad Auction Was Second-Price. The FTC Says You Paid Your Own Bid Anyway. (2026-09-10) | https://www.adbadger.com/blog/amazon-says-its-ad-auction-was-second-price-the-ftc-says-you-paid-your-own-bid-anyway/ | practitioner (fetched by the verifier) | 2026-11-09 | Price-paid versus who-wins distinction |
+| Intentwise, New controls for offsite Sponsored Products placements | https://www.intentwise.com/blog/new-controls-for-offsite-sponsored-products-placements | practitioner | 2026-11-09 | Off-Amazon controls; no site breakdown |
+| PPC Land, Amazon Ad Lab names four bidding metrics | https://ppc.land/amazon-ad-lab-names-four-bidding-metrics-but-publishes-no-test-results/ | practitioner press | 2026-11-09 | Ad Lab fixed-bid wording |
+| Novadata, ROAS auto-bidding 2026 | https://novadata.io/resources/news/amazon-sponsored-products-roas-auto-bidding-2026 | practitioner | 2026-11-09 | Unverified ROAS beta claim |
+| Amalyze, Amazon PPC bid range explained (added by the verifier) | https://amalyze.com/resources/guides/advertising/amazon-ppc-bid-range-explained | practitioner | 2026-11-09 | Backwards-looking summary; 25th–75th percentile reading |
+| Vendor pages on the suggested bid (Seller Metrics, FiveX, Feedvisor) | (search results, several URLs) | practitioner (search) | 2026-11-09 | Percentile and "high end" readings of the range |
+| Perpetua help and Intentwise blog on TOS share; Pacvue TOS post | (search results) | practitioner (search) | 2026-11-09 | Per-target TOS share; 10–15% rough figure; base-bid vs multiplier |
+| Chen, Nabi & Siniscalchi, Advancing Ad Auction Realism | https://cdn.amazon.science/a5/3f/f4a65055483ba954a1bc476a17fc/advancing-ad-auction-realism-practical-insights-modeling-implications.pdf | research (Amazon) | 2026-11-09 | Soft-floor rule and simulations |
+| amzn/auction-gym README | https://github.com/amzn/auction-gym | research (Amazon repo) | 2026-11-09 | Simulator scope |
+| A probabilistic framework to learn auction mechanisms via gradient descent | https://www.amazon.science/publications/a-probabilistic-framework-to-learn-auction-mechanisms-via-gradient-descent | research (Amazon) | 2026-11-09 | Publication record |
+| Balseiro et al., Robust Auction Design in the Auto-bidding World | https://arxiv.org/abs/2111.02468 | research | 2026-11-09 | Reserves and boosts with autobidders |
+| Blake, Nosko & Tadelis, NBER w20171 | https://www.nber.org/papers/w20171 | research | 2026-11-09 | eBay paid-search experiments |
+| arXiv abstract pages and Crossref records for the research items (added by the verifier) | https://arxiv.org/abs/2111.02468 ; https://api.crossref.org/works/10.1509/jmr.08.0468 | research (bibliographic) | 2026-11-09 | Abstract wording, author affiliations, volume and pages |
+| Edelman, Ostrovsky & Schwarz 2007; Ghose & Yang 2009; Jerath et al. 2011; Agarwal, Hosanagar & Smith 2011; Narayanan & Kalyanam 2015; Ursu 2018; Zeithammer 2019; Aggarwal et al. 2019; Ostrovsky & Schwarz 2023; Joo, Shi & Abhishek 2024 | (journal and repository pages surfaced by search; DOIs in *Published research*) | research (search) | 2026-11-09 | Abstract-level findings |
+| What agentic shopping means for advertising (gap fill) | https://advertising.amazon.com/library/news/agentic-shopping-advertising | official news | 2026-11-09 | Rufus and Alexa+ merged into Alexa for Shopping; automatic eligibility; prompts; Alexa+ Conversational Ads auto-enrolment; nothing on bids or placement adjustments |
+| Meet Alexa for Shopping, your personalized, agentic AI assistant on Amazon (gap fill) | https://www.aboutamazon.com/news/retail/alexa-for-shopping-ai-assistant | official corporate | 2026-11-09 | Merger announced 2026-05-13; US; search bar and product pages |
+| 'Amazon Rufus' AI experience comes to the Amazon Shopping app (gap fill) | https://www.aboutamazon.com/news/retail/amazon-rufus | official corporate | 2026-11-09 | "On May 13, 2026, Rufus was renamed Alexa for Shopping" |
+| Understand prompts on Sponsored Products and Sponsored Brands (re-read, gap fill) | https://advertising.amazon.com/help/GLWNW26EWQZW46WZ | official help | 2026-11-09 | CPC from existing budget; no placement or bid rule |
+| Prompt Ad Extension reports (gap fill) | https://advertising.amazon.com/API/docs/en-us/guides/reporting/v3/report-types/prompt-ad-extension | official API guide | 2026-11-09 | spPromptAdExtension, sbPromptAdExtension; US; no placement or bid column |
+| Placement classification (Supply), v1 dimension (re-read, gap fill) | https://advertising.amazon.com/API/docs/en-us/guides/reporting/ads-v1/dimensions/supply/placement-classification | official API guide | 2026-11-09 | No published value list |
+| Campaign management in the Amazon Ads API (overview, gap fill) | https://advertising.amazon.com/API/docs/en-us/guides/campaign-management/overview | official API guide | 2026-11-09 | Lists Sponsored Display as supported in v1 |
+| Release notes: Ads API v1 (gap fill) | https://advertising.amazon.com/API/docs/en-us/release-notes/ads-api-v1 | official API notes | 2026-11-09 | SD "later in 2025"/"later this year"; no SD GA note |
+| Amazon Ads API v1, Sponsored Products / Sponsored Brands / Sponsored Display models (gap fill) | https://d1y2lf8k3vrkfu.cloudfront.net/openapi/en-us/dest/AmazonAdsAPISP_prod_3p.json ; …/AmazonAdsAPISB_prod_3p.json ; …/AmazonAdsAPISD_prod_3p.json | official spec | 2026-11-09 | Per-product update ceilings; SP/SB bid strategy and placement enums; SD v1 bid and KPI fields |
+| Sponsored Products 3.0, v1 merged, Sponsored Brands 3.0 and 4.0, Sponsored Display 3.0 OpenAPI (re-read for writes, gap fill) | URLs as in the rows above | official spec | 2026-11-09 | Update operations, media types, `maxItems`, 207 shapes, error reasons, optimization-rule notes |
+| Campaign reports and Targeting reports (v3), Reporting columns (version 3) (re-read, gap fill) | URLs as in the rows above | official API guide | 2026-11-09 | Which report types and groupings list topOfSearchImpressionShare |
+| Sponsored Products performance datasets (re-read, gap fill) | https://advertising.amazon.com/API/docs/en-us/guides/amazon-marketing-stream/datasets/sp-performance | official API guide | 2026-11-09 | Dataset ids `sp-traffic` and `sp-conversion` |
+| Reserve keywords in a Sponsored Brands campaign (help G86SD7HK6NHHRB9B) (gap fill) | https://advertising.amazon.com/help/G86SD7HK6NHHRB9B | official help | 2026-11-09 | 15–92 days; $3,000 minimum; not in Spain |
+| How to set up Amazon Sponsored Brands reserve share of voice (gap fill) | https://advertising.amazon.com/library/guides/sponsored-brands-reserve-share-of-voice | official library | 2026-11-09 | e.g. 14–90 days; $6K US; 20 marketplaces incl. Ireland |
+| Campaign goals for Sponsored Brands; Bidding strategies for Sponsored Products; Set bids; Understand bid adjustments; Adjust Sponsored Products bids; Dynamic bidding for Sponsored Products; An author's guide; Sponsored Products best practices (re-read, gap fill) | URLs as in the rows above | official help and library | 2026-11-09 | Up-and-down ceiling wording; reserve SOV 30–92 days |
+| Crossref records: Rutz, Bucklin & Sonnier 2012; Agarwal et al. 2011; Ghose & Yang 2009; Narayanan & Kalyanam 2015; Reiley, Li & Lewis (SSRN 2190189; EC 2010); Craswell et al. 2008; Joachims et al. (WSDM 2017, IJCAI 2018) (gap fill) | https://api.crossref.org/works/10.1509/jmr.10.0354 (and the other DOIs given in *Published research*) | research (bibliographic, abstracts) | 2026-11-09 | Verbatim abstracts on conversion by position, north-ad externalities |
+| Craswell, Zoeter, Taylor & Ramsey, An experimental comparison of click position-bias models (gap fill) | https://www.microsoft.com/en-us/research/publication/an-experimental-comparison-of-click-position-bias-models/ | research | 2026-11-09 | Cascade model; abstract |
+| Joachims, Swaminathan & Schnabel, Unbiased Learning-to-Rank with Biased Feedback (gap fill) | https://arxiv.org/abs/1608.04468 | research | 2026-11-09 | Propensity-weighted LTR; abstract |
+| Xue et al., Practical Lessons on Optimizing Sponsored Products in eCommerce (gap fill) | https://arxiv.org/abs/2304.09107 (PDF https://arxiv.org/pdf/2304.09107) | research (Walmart) | 2026-11-09 | Walmart eCommerce affiliation; position-weighted training |
+| Wang et al., Click-Conversion Multi-Task Model with Position Bias Mitigation (gap fill) | https://arxiv.org/abs/2307.16060 | research | 2026-11-09 | UIC, Walmart eCommerce and Indeed affiliations; joint CTR/CVR de-biasing |
+| Karra, Zhao, Murray & Pellegrini, Nudging neural click prediction models to pay attention to position (gap fill) | https://www.amazon.science/publications/nudging-neural-click-prediction-models-to-pay-attention-to-position | research (Amazon) | 2026-11-09 | CIKM 2023; self-reinforcing position bias |
+| Lewis, MIT thesis abstract (gap fill) | https://dspace.mit.edu/handle/1721.1/62403 | research | 2026-11-09 | North-ads chapter; engine not named |
+
+### Unreachable
+
+- Several Amazon Ads help node ids (G7ZEXWWPDBZDZ8NZ, G2QUCQQ8A99X43B5, GBXUTNPZ2AXCD8KT, GD9YVKLWV4879NCN, GJ7FKQWK7ZKG5YDK, GMYGZ8D49STNTKAG, GSXM697WAXDBRHBM, GF2HQ7XLD3CVCA2X, GKVPWHKPLMPP6UJ7, GNFLB5F86PDRSNCY, GNWG4L67BKT7ZS99, GVMA56GGSEUM7UVB): the GET returned the shell without an embedded article for that id (probably category or redirect nodes). GPDRGYSKVB7H4BUR likewise returned no article.
+- aboutamazon.com `/news/...` address of the FTC response: 404 or a Next.js error shell; the `/md/` copy was read instead.
+- aboutamazon.com post on Sponsored Products with Pinterest and BuzzFeed: unreachable at both the `/news/` and `/md/` addresses.
+- Amazon Ads library guides *Bidding strategies*, *PPC advertising*, *What is PPC*, *vCPM*, *Viewable impressions*, *Sponsored Brands best practices*, *Sponsored Display best practices*: each address returned Amazon Ads' "404 Not Found" page (re-checked by the verifier on 2026-11-09).
+- Cloudfront markdown for `guides/sponsored-products/bidding`, `guides/sponsored-products/placement`, `guides/sponsored-products/bidding/shopper-cohort`, `guides/sponsored-brands/bidding` and `guides/sponsored-brands/bidding/overview`: `NoSuchKey`.
+- *Learning to Bid with AuctionGym* PDF (cdn.amazon.science): downloaded but not extracted in this session; the GitHub README and abstract were used.
+- Pacvue, "Three Reasons Winning Top of Search Doesn't Matter": HTTP 403.
+- Seller Metrics pages on the suggested bid: a plain GET returned no readable text; used through search-result summaries only.
+- Pulse (pulse.bot) item on "Smart Bid V2": the fetch failed (socket closed); used through a search-result summary only.
+- The FTC docket (FTC v. Amazon, W.D. Wash., 2:26-cv-03097) and the reported soft-reserve chronology (Sponsored Brands 2018, Sponsored Products 2019, display 2023): taken from search-result summaries; the complaint itself was not fetched.
+- m19 blog on bidding basics: HTTP 404.
+- The Amazon Ads MCP server's downloadable skills (`/skills` on `advertising-api*.amazon.com`): not fetched, because that host is an Ads API host this seat may not call.
+- Journal full texts behind INFORMS, AEA, JPE and Elsevier: not fetched; abstracts and summaries only.
+- The Amazon Ad Lab video on YouTube (2026-10-15): not watched; read through PPC Land.
+
+## Open questions
+
+1. Up-and-down ceiling: current help, the library guide and the API model say up to 100% for all placements; the author's guide and most practitioners say 100% top of search and 50% elsewhere. Which is live in each marketplace, and since when?
+2. Rule-based bidding: help says a ROAS target is optional and Amazon sets one if absent, and the cap is 5× the adjusted bid; the API guide requires a ROAS value and gives 10-day and 10-conversion eligibility. Are these two generations of the feature, or console versus API?
+3. Shopper-cohort recommendation units: `/sp/campaign/recommendations` describes the percentage as basis points ("900 = 9%"); the campaign field is a plain percent 0–900. Which unit does a recommended value carry?
+4. (Resolved by verification: the 3-entry maximum applies only to the target-promotion-group new-campaign shape and to new-ad-group bid recommendation requests; campaign create and update set no limit on `placementBidding`.)
+5. Which bid applies to prompt, Alexa for Shopping (formerly Rufus) and Alexa+ Conversational Ads clicks, and do any placement adjustments apply there? Re-read 2026-11-09 after the May 2026 merger: the pages say campaigns are automatically eligible or enrolled, prompts are CPC from the campaign budget and appear in existing reporting, and measurement is "closed-loop"; none says how they are priced against adjustments, nor whether the 2024 line "Your campaign reports will not include Rufus metrics" still holds for Alexa for Shopping.
+6. Schedule bid rules add to placement adjustments (help example), while placement and audience adjustments multiply. How does a schedule rule combine with an audience boost and with up-and-down bidding?
+7. The v1 reporting API drops top-of-search impression share below campaign level. Will a target-level share return before v3 reporting ends (2027-06-30, confirmed on the Deprecations page)? The Stream v2 `ads-performance-v1` dataset carries none either.
+8. Sponsored Brands negative placement adjustments were deprecated from 2026-06-15; what does the API now return for a negative value on create or update, and do existing negative values still show in reads?
+9. The FTC case: will a ruling or settlement change how reserves are applied or disclosed? Nothing here is settled as of 2026-11-09.
+10. (Resolved by verification: the release notes of 2024-08-08 and 2024-08-22 and the Deprecations page confirm the 2025-05-15 shutoff of `/v2/sp/targets/bidRecommendations` and of the v3 theme-based content type.)
+11. Novadata's "Automatically optimizing ROAS within budget" beta, and Pulse's reported "Smart Bid V2" (June 2026): no Amazon page read here describes either. What does the v1 `PRIORITIZE_KPI_TARGET` strategy do for Sponsored Products, if anything?
+12. What does the Sponsored Products 3.0 `autoManageCampaign` flag change about bids?
+13. A bid write to a display target whose ad group has an enabled optimization rule, or to a Sponsored Brands keyword or target in a campaign with one: is it refused with an item error or accepted and overridden? Amazon's models say "only mutable when … does not have any enabled optimization rule" and "manual bids for individual targets will be overridden", which leaves both readings open.
+14. The SD v3 model gives `PUT /sd/targets` two ceilings (100 in the operation text, 1,000 in the request body text); the per-product SP v1 model allows one audience adjustment per campaign where the merged v1 model allows 10; the v1 `ErrorsIndex.index` is bounded at 19. Which are live?
+15. Is Sponsored Display on v1 campaign management generally available? The overview lists it and a model exists, but no release note dates it (`method/ads-api-surface.md` holds the field reports).
+
+## Positions
+
+**P-bids-and-placements-1.** On Amazon a bid is the maximum the advertiser authorizes per click (or per 1,000 viewable impressions); Amazon states it never charges more than the maximum effective bid, which includes bid adjustments and the bidding strategy. Sources: [official] *Understand Amazon Ads auctions* (help GEXGZNBBWPZZK7P8); [official] *Understand bidding on Sponsored ads* (help GDS2E64R9GQ35768); [official] *What is cost per click (CPC)?*; all read 2026-11-09.
+
+**P-bids-and-placements-2.** Amazon describes its sponsored-ads auction as "a form of generalized “second-price” auction" with a hard reserve that a bid must exceed to enter, and soft reserves; a winner above both pays the soft reserve, and a winner above the hard but below the soft reserve still wins and pays its own bid. Sources: [official] *Amazon's response to the FTC's lawsuit regarding Sponsored Ads* (aboutamazon.com); [official] *Understand Amazon Ads auctions*; read 2026-11-09.
+
+**P-bids-and-placements-3.** Amazon states that "the price charged to you may exceed the runner-up bid", and that reserves move with factors including predicted sale likelihood, predicted return on ad spend, competing bids, placement, context and peak periods. Sources: [official] *Understand Amazon Ads auctions*; [official] *Understand auctions for Sponsored Products ads across retailers* (help GW3CP75ZUTET4QEF); read 2026-11-09.
+
+**P-bids-and-placements-4.** Amazon ranks ads by relevance and bid, and states that ads with lower bids and higher relevance "can and do win"; it says that in 2024 about 92% of selected Sponsored Products ads were not the highest bid (figure for human readers only). Sources: [official] *Understand Amazon Ads auctions*; [official] *Amazon's response to the FTC's lawsuit*; read 2026-11-09.
+
+**P-bids-and-placements-5.** The FTC and 22 states allege (complaint of 2026-08-31, untested) that Sponsored Products advertisers paid their own bid 30–40% of the time in 2021, 70% in 2022 and about 80% in 2024 because of an undisclosed soft reserve (figures for human readers only). Sources: [official] *Amazon's response to the FTC's lawsuit regarding Sponsored Ads* (disputes the characterisation), read 2026-11-09; [opinion: FTC, regulator] *FTC, States Sue Amazon Over Secret Ad Surcharge Scheme*, read 2026-11-09.
+
+**P-bids-and-placements-6.** Sponsored Products offers four bidding strategies, named in the 3.0 API `LEGACY_FOR_SALES` (dynamic bids, down only), `AUTO_FOR_SALES` (up and down), `MANUAL` (fixed) and `RULE_BASED`; a campaign created through the API without a strategy gets `LEGACY_FOR_SALES`. Sources: [official] Sponsored Products 3.0 OpenAPI; [official] *Bidding strategies for Sponsored Products* (help GCU2BUWJH2W3A8Z7); read 2026-11-09.
+
+**P-bids-and-placements-7.** Under "Dynamic bids - down only" Amazon may lower a bid by up to 100% for clicks it judges less likely to convert, and never raises it. Sources: [official] *Bidding strategies for Sponsored Products*; [official] Sponsored Products 3.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-8.** Under "Dynamic bids - up and down" Amazon's current help, library guide and Sponsored Products 3.0 and v1 API models say it may raise or lower a bid by up to 100% for all placements ($1.00 up to $2.00), while Amazon's author's guide still says up to 100% for top of the first page and up to 50% for other placements. Sources: [official] *Bidding strategies for Sponsored Products*; [official] *Dynamic bidding for Sponsored Products*; [official] Sponsored Products 3.0 OpenAPI; [official] Amazon Ads API v1 merged OpenAPI; [official] *An author's guide to Sponsored Products*; read 2026-11-09.
+
+**P-bids-and-placements-9.** Under fixed bids Amazon uses the exact bid plus any placement adjustments and does not apply dynamic bidding; Amazon says fixed bids may give more impressions but fewer conversions for the spend. Sources: [official] *Bidding strategies for Sponsored Products*; [official] Sponsored Products 3.0 OpenAPI (`MANUAL`); read 2026-11-09.
+
+**P-bids-and-placements-10.** Sponsored Products placement adjustments are increases of 0% to 900% for top of search (first page: up to the first 5 positions of the top row on desktop, 4 on mobile), rest of search, product pages and Amazon Business placements. Sources: [official] *Understand bid adjustments in Sponsored ads* (help GQ4R4FR4H56JL6V8); [official] *Adjust Sponsored Products bids* (help GYYZVM7LGSRYGWV5); [official] Sponsored Products 3.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-11.** Dynamic bidding applies to the placement-adjusted bid: Amazon's example is a $1 bid with +50% top of search giving $1.50, which up and down "can further increase ... up to $3". Sources: [official] *Understand bid adjustments in Sponsored ads*; [official] *Adjust Sponsored Products bids*; read 2026-11-09.
+
+**P-bids-and-placements-12.** In Sponsored Products, adjustments of different kinds multiply (placement × audience × video), and when a shopper is in several boosted audiences only the highest audience boost applies, with ties attributed to one audience at random. Sources: [official] *Adjust Sponsored Products bids*; [official] *Understand bid adjustments in Sponsored ads*; [official] Sponsored Products 3.0 OpenAPI (`shopperCohortBidding`); read 2026-11-09.
+
+**P-bids-and-placements-13.** Sponsored Products allows up to 10 boosted audiences per campaign (Amazon audiences or AMC custom audiences), each 0% to 900%; outside the US an advertiser can select only 1 audience. Sources: [official] *Bidding strategies for Sponsored Products*; [official] *Set bids in a Sponsored Products campaign* (help GTMXQWASBRZTHLF2); [official] Sponsored Products 3.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-14.** Sponsored Products off-Amazon placements use the campaign's own bid and budget; top-of-search, rest-of-search and product-page adjustments do not apply to them, while dynamic and rule-based bidding do; the API switch is `offAmazonBudgetControlStrategy` `MAXIMIZE_REACH` (default in the console) or `MINIMIZE_SPEND`. Sources: [official] *Understand Sponsored Products off-Amazon advertising* (help GYTD2Z3SYMAAMVXA); [official] Sponsored Products 3.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-15.** Off Amazon, Amazon adjusts the bid on third-party sites by predicted click-through and conversion and submits it to the publisher, who decides which ads show; for creator content Amazon sets the CPC from the maximum bid, expected performance and the sales the ad generates. Sources: [official] *Understand Sponsored Products off-Amazon advertising*; read 2026-11-09.
+
+**P-bids-and-placements-16.** A site restriction (Amazon Business only, or Amazon Haul in the US) cannot be changed after the campaign is created, and such a campaign cannot use audience (shopper-cohort) bid adjustments or the Amazon Business placement adjustment. Sources: [official] *Sites selection for Sponsored Ads campaigns* (help GJGYD2FKNF3MUWBQ); [official] Sponsored Products 3.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-17.** Rule-based bidding adjusts base bids per impression toward a ROAS goal, can raise bids up to 5 times the adjusted bid, and carries no guarantee of the goal; Amazon's help says that without a ROAS target Amazon sets one. Sources: [official] *Bidding rules for Sponsored Products* (help GDLQ5D2BNFCAU3TW); [official] *Sponsored Products Rule-based bidding overview*; read 2026-11-09.
+
+**P-bids-and-placements-18.** Amazon's API guide states that a rule-based bidding rule is switched off automatically when 20-day ROAS falls below 30% of ROAS over the 20-to-40-day window, and the campaign's earlier strategy and bids return. Sources: [official] *Sponsored Products Rule-based bidding overview*; read 2026-11-09.
+
+**P-bids-and-placements-19.** Schedule bid rules cannot be applied to a campaign with rule-based bidding active, and go on hold if the campaign switches to rule-based bidding. Sources: [official] *Bidding rules for Sponsored Products*; read 2026-11-09.
+
+**P-bids-and-placements-20.** Schedule bid rules are Sponsored Products only and only increase bids, by 1% to 100% of the base bid, for an hour window, weekday, date range or event; several rules add together, and the total adds to the placement adjustment (Amazon's example: $1.00, top of search +10%, Monday rule +20% gives $1.30). Sources: [official] *Bidding rules for Sponsored Products*; [official] *Scheduled bid rules* (API guide); read 2026-11-09.
+
+**P-bids-and-placements-21.** Amazon states that a schedule bid rule change takes effect within 24 hours and that a campaign may carry at most 20 schedule bid rules. Sources: [official] *Scheduled bid rules* (API guide); read 2026-11-09.
+
+**P-bids-and-placements-22.** Amazon says suggested bids and bid ranges are calculated from bids of similar ads that won impressions over the past 7 days and are updated daily; for events in the next 45 days the console shows a separate event suggestion. Sources: [official] *Understand bidding on Sponsored ads*; [official] *Set bids in a Sponsored Products campaign*; [official] Sponsored Display 3.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-23.** Amazon states that Performance Estimate figures are not a forecast of campaign performance and do not consider budget constraints; the quick-start guide says the same of the clicks-and-orders impact metrics ("not estimates of expected performance", computed assuming the campaign does not run out of budget), which belonged to the v3 theme-based content type shut off on 2025-05-15 and were removed in v4. Sources: [official] *Performance Estimate (beta)* (help GHFYQ2D4LQC5ABPW); [official] *Theme-based bid suggestions - Quick-start guide*; [official] *Amazon Ads advanced tools release notes* (2024-08-22) and *Deprecations*; read 2026-11-09.
+
+**P-bids-and-placements-24.** `POST /sp/targets/bid/recommendations` returns up to 3 bid values per target by theme (`CONVERSION_OPPORTUNITIES`, `PRIME_DAY`, `FALL_PRIME_DEAL_EVENT`, `BFCM_HOLIDAY`); the v5 content type adds estimated impressions for the suggested bid and, with `includeAnalysis`, impression estimates per placement for eight bid values, which Amazon says is currently available only for existing ad group requests in the US; an unsupported marketplace or target returns 422. Sources: [official] Sponsored Products 3.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-25.** Amazon defines top-of-search impression share as top-of-search impressions earned over the top-of-search impressions the campaign was eligible for; for Sponsored Products it covers the first page. Sources: [official] *Reporting columns (version 3)* (reporting v3); [official] *Top-of-search impression share metric* (what's new, 2021-09-22); read 2026-11-09.
+
+**P-bids-and-placements-26.** In the v3 reporting API, top-of-search impression share is listed for `spCampaigns`, `sbCampaigns` and `sbTargeting` on the columns page, for `spTargeting` and `sbTargeting` base metrics on the targeting report page, and for `spCampaigns` grouped by `campaign` or `campaignPlacement` on the campaign report page; in the v1 reporting API it is compatible with campaign-level dimensions only, not with target, search term, ad group, hour or placement. Sources: [official] *Reporting columns (version 3)*, *Targeting reports* and *Campaign reports* (v3); [official] v1 *Search (Impression share)* metrics page; read 2026-11-09.
+
+**P-bids-and-placements-27.** Amazon's v3 reporting has no separate Sponsored Products placement report: placement performance comes from `spCampaigns` grouped by `campaignPlacement`, and off-Amazon and creator traffic appear there as "Off Amazon" or "Other – On Amazon". Sources: [official] *Placement reports* (v3); [official] *Understand Sponsored Products off-Amazon advertising*; read 2026-11-09.
+
+**P-bids-and-placements-28.** In Sponsored Brands only top of search and rest of search (placements other than top of search) can be adjusted; automated bidding lets Amazon optimize placements below top of search, and in the 4.0 API `bidOptimization: true` makes Amazon ignore `bidAdjustmentsByPlacement`. Sources: [official] *Understand bid adjustments in Sponsored ads*; [official] *Create a Sponsored Brands campaign* (help GF86HBCNDJUAC5WN); [official] Sponsored Brands 4.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-29.** Amazon's release notes state that new negative Sponsored Brands placement adjustments are not accepted from 2026-06-15 (first announced for 2026-02-02, extended on 2026-01-27), while existing ones keep running with their current settings; positive adjustments go up to 900% for top of search and rest of search (announced 2025-06-19; rest of search was limited to 99% before). Sources: [official] *Amazon Ads advanced tools release notes* (2025-06-19, 2025-12-01, 2026-01-27); read 2026-11-09.
+
+**P-bids-and-placements-30.** For Sponsored Brands campaigns with the goal "drive page visits", Amazon may raise a bid by up to 300% for an auction and lower it where conversion looks less likely. Sources: [official] *Understand bidding on Sponsored ads*; [official] Ads API release notes (2023-10-05); read 2026-11-09.
+
+**P-bids-and-placements-31.** Sponsored Brands and display cost control set a target average CPC, not a maximum: Amazon states the average CPC may be higher than the amount entered and that with placement or audience adjustments the highest possible CPC exceeds the target. In Sponsored Brands cost control is on automatically for the goal "drive page visits" and only available there, and turning it off means entering bids; in display, bids are set automatically and cannot be edited by hand, and suggested bids are not shown. Sources: [official] *Understand Cost controls for sponsored ads* (help GNJDAT6AEYD3XJA8); [official] *Sponsored Brands bids by placement and audiences* (help GC8K522SMNCPH99Z); read 2026-11-09.
+
+**P-bids-and-placements-32.** Sponsored Brands allows one audience bid adjustment per campaign (Amazon-built or AMC, not both), and a campaign's goal cannot be changed after creation. Sources: [official] *Sponsored Brands bids by placement and audiences*; [official] *Campaign goals for Sponsored Brands* (help GXXPFXLSVX9VD3U9); [official] Sponsored Brands 4.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-33.** Reserve share of voice gives Sponsored Brands a fixed, upfront price for top-of-search placement on 5 to 1,000 branded keywords, outside the auction, and Amazon says it strives for 95% share of voice; Amazon's pages give different flights (30–92 days on *Campaign goals for Sponsored Brands*, 15–92 on *Reserve keywords in a Sponsored Brands campaign*, "e.g., 14–90" on the library guide), different US minimums ($3,000 or $6K) and different marketplace lists (two help pages exclude Spain; the library guide lists 20 marketplaces including Ireland), as `beyond/brand-defense-and-competition.md` records in full (figures for human readers only). Sources: [official] *Campaign goals for Sponsored Brands* (help GXXPFXLSVX9VD3U9); [official] *Reserve keywords in a Sponsored Brands campaign* (help G86SD7HK6NHHRB9B); [official] *How to set up Amazon Sponsored Brands reserve share of voice*; [official] *Reserve share of voice* (Sponsored Brands API guide); read 2026-11-09.
+
+**P-bids-and-placements-34.** In display ads (formerly Sponsored Display) Amazon may raise a bid by up to 300% for an auction under any optimization strategy ($1 to $4); the strategy (reach on vCPM; page visits, conversions or leads on CPC) cannot be changed after creation. Sources: [official] *Optimization strategies for Display* (help GDRFG6U7TRD72EFW); [official] *Amazon Ads API for Sponsored Display overview* (API guide); read 2026-11-09.
+
+**P-bids-and-placements-35.** For vCPM, the bid is deducted per 1,000 viewable impressions, a view following MRC (at least 50% of the ad in view for at least 1 second); Amazon publishes per-country vCPM minimums, and its display help warns that a vCPM bid below $3 "may not lead to viewable impressions" (figure for human readers only). Sources: [official] *Understand bidding on Sponsored ads*; [official] *Limits, constraints, and quotas*; [official] *Optimization strategies for Display* (help GDRFG6U7TRD72EFW); read 2026-11-09.
+
+**P-bids-and-placements-36.** Amazon states that click data reach the API within about 12 hours, that clicks can be invalidated for 72 hours afterwards, and that Sponsored Products conversions in the Marketing Stream v1 dataset can be revised for up to sixty days after the click; Stream Version 1 stops delivering data on 2027-06-30, and the Stream v2 `ads-performance-v1` dataset states a 70-day restatement window. Sources: [official] *Limits, constraints, and quotas*; [official] *Sponsored Products performance datasets* (Marketing Stream); [official] *Deprecations*; [official] *Ads performance* (Stream v2); read 2026-11-09.
+
+**P-bids-and-placements-37.** BidX reports, from about 130,000 campaigns in February 2024, that down-only campaigns had the lowest ACOS and better conversion than up-and-down despite higher CPCs, and that within down-only campaigns a product-page adjustment above 0% went with the worst ACOS; the data are observational. Sources: [opinion: BidX, Lena Einschütz] *Case study on the dynamic bidding strategies of Amazon 2024*, read 2026-11-09.
+
+**P-bids-and-placements-38.** Perpetua reports, from six weeks of US Sponsored Products data in spring 2022, average CPC falling from $1.13 to $0.84 over the day and ACOS rising from 20–23% in the daytime to 41% at 11pm (figures for human readers only). Sources: [opinion: Perpetua, Stephen Bench-Capon] *Study: Intraday Optimization*, read 2026-11-09.
+
+**P-bids-and-placements-39.** Narayanan and Kalyanam find, with a regression-discontinuity design, that naive estimates overstate position effects in search advertising and that the effects are larger for smaller advertisers and less experienced consumers; Agarwal, Hosanagar and Smith find that profit can be higher below the top position. Sources: [research] Narayanan & Kalyanam, *Marketing Science* 2015; [research] Agarwal, Hosanagar & Smith, *JMR* 2011; abstracts read 2026-11-09.
+
+**P-bids-and-placements-40.** Researchers at Amazon Ads and Outreach describe a soft floor under which a bidder below the floor still wins and pays its own bid, and find in simulation that soft floors barely change revenue with symmetric single-query bidders but can change it when bidders choose which queries to target. Sources: [research] Chen (Outreach Inc.), Nabi & Siniscalchi (Amazon Ads), *Advancing Ad Auction Realism*, AdKDD 2023; [research] Zeithammer, *Management Science* 2019; read 2026-11-09.
+
+**P-bids-and-placements-41.** Amazon merged Rufus and Alexa+ into one shopping assistant, Alexa for Shopping, announced on 2026-05-13 for US customers in the Amazon app, the website and Echo Show; Rufus "was renamed Alexa for Shopping". Sources: [official] *Meet Alexa for Shopping, your personalized, agentic AI assistant on Amazon* (aboutamazon.com); [official] *'Amazon Rufus' AI experience comes to the Amazon Shopping app* (aboutamazon.com); [official] *What agentic shopping means for advertising* (Amazon Ads news, 2026-06-11); read 2026-11-09.
+
+**P-bids-and-placements-42.** Amazon states that existing sponsored ads campaigns are automatically eligible to serve in Alexa for Shopping and to be optimized as Sponsored Products and Sponsored Brands prompts, and that advertisers with sponsored ads campaigns are automatically enrolled in Alexa+ Conversational Ads on Alexa-enabled devices. Sources: [official] *What agentic shopping means for advertising* (2026-06-11); [official] *Understand prompts on Sponsored Products and Sponsored Brands* (help GLWNW26EWQZW46WZ); read 2026-11-09.
+
+**P-bids-and-placements-43.** No Amazon page read on 2026-11-09 says which bid adjustments (placement, audience, video) or bidding strategy apply to ads in Alexa for Shopping, prompts or Alexa+ Conversational Ads; Amazon says prompts are charged per click from the existing campaign budget, and its prompt reports carry cost and conversions but no placement or bid column. Sources: [official] *What agentic shopping means for advertising*; [official] *Understand prompts on Sponsored Products and Sponsored Brands*; [official] *Prompt Ad Extension reports* (v3); [official] Ads API release notes; read 2026-11-09.
+
+**P-bids-and-placements-44.** In the Sponsored Products 3.0 API, bids are written with `PUT /sp/keywords` (`keywordId`, `bid`), `PUT /sp/targets` (`targetId`, `bid`), `PUT /sp/adGroups` (`adGroupId`, `defaultBid`) and `PUT /sp/campaigns` (`dynamicBidding.strategy`, `placementBidding[]`, `shopperCohortBidding[]`), each with its own vendor media type (`application/vnd.spKeyword.v3+json`, `vnd.spTargetingClause.v3+json`, `vnd.spAdGroup.v3+json`, `vnd.spCampaign.v3+json`) and at most 1,000 items per call (figure for human readers only). Sources: [official] Sponsored Products 3.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-45.** A Sponsored Products 3.0 bid write answers 207 with `success` and `error` lists keyed by request index; a refused bid carries a `biddingError` with a reason (`BID_OUT_OF_MARKET_PLACE_RANGE`, `BID_GT_BUDGET`, `BID_INVALID_PLACEMENT`, or an audience reason) and the marketplace's `lowerLimit` and `upperLimit`, and an archived or parent-archived entity is refused with `entityStateError` reasons such as `ARCHIVED_ENTITY_CANNOT_BE_MODIFIED` and `PARENT_ARCHIVED_FORBIDS_UPDATES`. Sources: [official] Sponsored Products 3.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-46.** In the unified v1 API, bids are written with `POST /adsApi/v1/update/targets` (`bid.bid`), `update/adGroups` (`bid.defaultBid`) and `update/campaigns` (`optimizations.bidSettings.bidStrategy` and `bidAdjustments`) in `application/json`; per-call ceilings differ by ad product (Sponsored Products 1,000; Sponsored Brands 10 campaigns or ad groups and 1,000 targets; Sponsored Display 100 campaigns or ad groups and 1,000 targets; figures for human readers only), and refusals carry codes such as `FIELD_VALUE_IS_OUT_OF_RANGE`, `ARCHIVED_RESOURCE_CANNOT_EDIT` and `AUTOCREATED_ENTITY_CANNOT_EDIT`. Sources: [official] Amazon Ads API v1 merged OpenAPI; [official] *Amazon Ads API v1, Sponsored Products / Sponsored Brands / Sponsored Display* models; read 2026-11-09.
+
+**P-bids-and-placements-47.** Sponsored Brands keyword and target bids are written with `PUT /sb/keywords` and `PUT /sb/targets` (at most 100 per call; keywords require `keywordId`, `adGroupId` and `campaignId`), and placement and audience adjustments with `PUT /sb/v4/campaigns` (`bidding.bidOptimization`, `bidAdjustmentsByPlacement`, `shopperCohortBidAdjustments`; at most 10 campaigns per call); Amazon's model says a keyword or target bid "is only mutable when the keyword's corresponding campaign does not have any enabled optimization rule". Sources: [official] Sponsored Brands 3.0 OpenAPI; [official] Sponsored Brands 4.0 OpenAPI; read 2026-11-09.
+
+**P-bids-and-placements-48.** Sponsored Display bids are written with `PUT /sd/targets` (`bid`) and `PUT /sd/adGroups` (`defaultBid`); Amazon's model says a target's bid is "only mutable when the targeting clause's adGroup does not have any enabled optimization rule" and that "When an optimization rule is associated to an ad group, manual bids for individual targets will be overridden", and the display cost-control help says "Bids can’t be manually edited". Sources: [official] Sponsored Display 3.0 OpenAPI; [official] *Understand Cost controls for sponsored ads* (help GNJDAT6AEYD3XJA8); read 2026-11-09.
+
+**P-bids-and-placements-49.** Amazon's campaign-management overview lists Sponsored Display among the products the v1 campaign API "currently supports", and Amazon publishes a Sponsored Display v1 model with update operations, while no release note read on 2026-11-09 dates Sponsored Display's general availability on v1. Sources: [official] *Campaign management in the Amazon Ads API*; [official] *Amazon Ads API v1, Sponsored Display* model; [official] *Release notes: Ads API v1*; read 2026-11-09.
+
+**P-bids-and-placements-50.** The Stream Version 1 Sponsored Products datasets are subscribed to by the ids `sp-traffic` (clicks, impressions, cost) and `sp-conversion` (conversions); both carry `keyword_id`, `placement` and an hourly `time_window_start`, and Stream Version 1 stops delivering data on 2027-06-30. Sources: [official] *Sponsored Products performance datasets* (Marketing Stream); [official] *Deprecations*; read 2026-11-09.
+
+**P-bids-and-placements-51.** Published studies disagree on how conversion rate moves with ad position: Ghose and Yang find it highest at the top, Rutz, Bucklin and Sonnier find higher positions raise both click-through and conversion, Agarwal, Hosanagar and Smith find conversion rate rising lower down, and Narayanan and Kalyanam show that auction-set positions make naive position comparisons biased by selection and find smaller causal effects at some positions only. Sources: [research] Ghose & Yang, *Management Science* 2009; [research] Rutz, Bucklin & Sonnier, *JMR* 2012; [research] Agarwal, Hosanagar & Smith, *JMR* 2011; [research] Narayanan & Kalyanam, *Marketing Science* 2015; abstracts read 2026-11-09.
+
+**P-bids-and-placements-52.** Click-model research finds that the probability of a click depends on display position independently of relevance, best explained in early ranks by a cascade in which users scan top to bottom and stop at the first worthwhile result, and corrects for it by modelling position or weighting by its propensity; a field experiment found that more rival ads above the organic results raised, rather than lowered, the top sponsored listing's click-through rate. Sources: [research] Craswell, Zoeter, Taylor & Ramsey, WSDM 2008; [research] Joachims, Swaminathan & Schnabel, WSDM 2017; [research] Reiley, Li & Lewis, ACM EC 2010; read 2026-11-09.
+
+**P-bids-and-placements-53.** Sponsored-products systems at Walmart (Xue et al. 2023; Wang et al. 2023) and Amazon researchers (Karra et al., CIKM 2023) treat position as a confounder of click and conversion data and de-bias their models for it; Xue et al. state that "more clicks on an ad item is not necessarily equal to the preference of this item from users, it may be just because this ad item is in a good position". Sources: [research] Xue, Liu, Du, Korlimarla & Wen, arXiv 2304.09107; [research] Wang et al., arXiv 2307.16060; [research] Karra, Zhao, Murray & Pellegrini, amazon.science; read 2026-11-09.
+
+### Seat's positions (enter no skill)
+
+- [seat] The full chain of an adjusted Sponsored Products bid, assembled from separate Amazon examples, is base × (1 + placement) × (1 + top audience) × (1 + video), then the strategy (down only to zero, up and down to ×2, rule-based to ×5), with schedule rules adding to the placement percentage. No single Amazon page states the chain.
+- [seat] With soft reserves, the price paid sits nearer the bid than the "second price plus a cent" folklore suggests, so a bid set by revenue per click × target ACOS behaves closer to a cost than to a ceiling.
+- [seat] Placement benchmarks that show top of search converting best (Perpetua, BidX) are selected comparisons; the research says naive position comparisons overstate the effect.
+- [seat] Amazon's native schedule surface can only add to bids, so a practitioner's "lower bids at night" is a direct bid write on a clock, not an Amazon feature.
+- [seat] Raw CTR and CVR by placement in the `campaignPlacement` report confound the placement with position bias, with the shoppers and queries Amazon's ranking sends there, and with the rival ads beside it; a placement multiplier judged on them is a biased comparison of unknown sign.
+- [seat] A bid write's 207 success shows the item was accepted, not that the auction uses the number: under an enabled optimization rule, or with an SD v1 write of the kind developers report as silently ignored, only a read-back and the next data show the effect.
+
+## Verification
+
+Verified 2026-11-09 by an adversarial verifier seat.
+
+Method. The file was read whole and every cited page was re-opened, not a sample: Amazon Ads help articles by plain GET (article text taken from the embedded `nodeId`/`content` JSON of the page, never through the help hub's content call); Ads API guides, release notes and the Deprecations page as markdown from the cloudfront host; the Sponsored Products 3.0, v1 merged, Sponsored Brands 3.0 and 4.0, Sponsored Display 3.0, Recommendations and MCP static specification files from their public cloudfront addresses, searched by schema name; library guides, product pages and what's new posts by GET; every practitioner page by GET (the FTC release and the Ad Badger FTC post through a fetch tool when a plain GET returned a shell); research items through their PDFs, abstract pages, publication pages and Crossref records. Each claim was checked for the page saying it (verbatim where quoted), the exact page name, the tag, whether the figure is current or superseded, and practitioner attribution. Web searches looked for 2025–2026 Amazon announcements that overturn a claim (up-and-down ceiling, Sponsored Brands placement ranges, theme-based recommendation versions, reporting and Stream deprecations, Sponsored Display in v1, rule-based bidding, a target-ROAS strategy, the FTC case). Positions were checked to state facts only. No Ads API host (`advertising-api*.amazon.com`), Login with Amazon, Amazon Ads MCP server endpoint, SP-API host or signed-in page was called, and no seller data was read.
+
+Corrections (was → now):
+
+1. A, winning bid: "fell about 50% from 2019" → the response's body says 2019 to 2024 and its key takeaways say 2019 to 2025.
+2. A, FTC: added that the year percentages are Sponsored Products figures, that a plain GET of the press release returns a "Page Not Found" shell, and that the docket and soft-reserve chronology come from search summaries; no ruling or settlement found.
+3. B, vCPM minimums: "Display 1 in most currencies (BR 2, EG 5)" → exceptions BR 2, EG 5, IN 4, JP 100, MX 5, SA 4, SG 4, TR 1.85.
+4. C, up-and-down ceiling note: "Scale Insights likewise" removed (its page states no ceiling); Trellis's 100/50 is attributed "to a third-party summary"; PPC Land's statement and the v1 model's `SALES_UP_AND_DOWN` / `SALES_DOWN_ONLY` wording added.
+5. C, "Wait about two weeks before judging a change" → the guide's wait is after a campaign's launch before more advanced bidding, then a check every two weeks.
+6. D, "a new campaign may carry at most 3 placement entries" → campaign create and update set no limit; the 3-entry maximum belongs to the target-promotion-group new-campaign shape and to new-ad-group bid recommendation requests (Open question 4 resolved).
+7. D, rest of search: release-note page title, 2023-09-21 beta date and 2023-08-06 preview added; the what's new page is dated 2024-01-09 and its full title given.
+8. E, video adjustment: "Increases apply on top of placement adjustments" added; the worked chain re-sourced; `creativeBidAdjustments` limits (2 entries, 0–900) added.
+9. E, v1 adjustments: the object is `BidAdjustments` inside `BidSettings`; `shopperSegmentBidAdjustments` at most 2, `creativeBidAdjustments` at most 2.
+10. E, boost reporting: title "Bid boosting for Sponsored Products and Sponsored Brands" → *Sponsored Products & Brands: Audience Bid Boosting*; "groups a campaign report" → the `spAudiences` and `sbAudiences` audience reports grouped by `campaign_bid_boost_segment`.
+11. F, Rufus: region `US` added.
+12. F, seat note "clicks from Rufus and Alexa+ are priced from the same bids" → the notes say nothing about pricing against adjustments; only reporting is stated.
+13. G, rule-based eligibility: "10 conversions in 30 days" → "in the last 30 days"; guardrail and recommendation sentences now verbatim ("based on the previous 21 days"; "historical campaign information and a group of similar advertisers").
+14. G, getting-started guide: content type, "Recommended refresh frequency is once a day" and "ROAS is the only supported metric" were attributed to the guide → they are in the Sponsored Products 3.0 OpenAPI (whose enum also lists `AVERAGE_BID`); page names corrected to *Sponsored Products Rule-based bidding overview* and *Getting started with the Sponsored Products rule-based bidding API*.
+15. G, target ROAS: v1 `PRIORITIZE_KPI_TARGET` and `SPEND_BUDGET_IN_FULL` descriptions quoted; Novadata's label "within the campaign budget" → "within budget".
+16. H, schedule rules: the 1–100 value wording, `INCREMENT`-only operator, `timesOfDay` limit and the v2 `PERFORMANCE` subcategory were attributed to the guide → they are in the Sponsored Products 3.0 OpenAPI; guide title *Schedule bid rules* → *Scheduled bid rules*; the 24-hour quote completed.
+17. I, versions: "v5 adds impact metrics" → v4 "removes 'impact metrics'", v5 adds "estimated impressions for the suggested bid"; v4 product-target types and `KEYWORD_GROUP` (US) spelled out.
+18. I, impact metrics: marked as v3-era; the v3 content type and the v2 endpoint shut off 2025-05-15 (notes of 2024-08-22 and 2024-08-08, Deprecations page) added; the 5-keyword minimum applies to keyword targeting only (Open question 10 resolved).
+19. J, v1 time zone: "per the advertiser's own time zone" → "aggregated based on each campaign timezone".
+20. J, v3 end date: "per the Ads API door reference" → confirmed on the Deprecations page (2027-06-30, with Stream v1 and Sponsored Ads v2 reporting); Stream v2 `ads-performance-v1` carries no impression share.
+21. K, Sponsored Brands ranges: the 2025-12-01 note set 2026-02-02, extended on 2026-01-27 to 2026-06-15; the 2025-06-19 wording quoted (a later note dates the top-of-search change 6/17/2025).
+22. K, Sponsored Brands cost control: "bids cannot be edited by hand, suggested bids are not shown" were the help page's Display bullets → the Sponsored Brands section's own statements (automatic for drive page visits, only there, bids needed when off, 7-day rolling average) replace them.
+23. K, Sponsored Brands optimization rules: the rule description and landing-page limit re-sourced to the Sponsored Brands 4.0 OpenAPI; "performance is evaluated over 21 days" → the guide's recommendation to adjust target or budget if not met over 21 days; beta status and "may add targets" added.
+24. K, goals: "goal, cost type and smart default cannot be edited" and BRAND_IMPRESSION_SHARE → VCPM re-sourced to the *Getting started with campaigns* guide; `FIXED_PRICE` is defined in the spec by `targetedPGDealId`; Ireland wording quoted.
+25. L, naming: "the old DSP products 'Display, Video, and Audio campaigns'" → the product page says existing Sponsored Display campaigns are reached through Display, Video, and Audio campaigns in Amazon Ads Agent.
+26. L, display cost control: rule overrides manual target bids and one rule per ad group added; US thresholds re-sourced to the Limits page (the spec does not carry them).
+27. L, v1: "come later" → "later this year" in a 2025 note, with no announcement found by 2026-11-09.
+28. M, Stream: `sp-performance` identified as a Stream v1 dataset ending 2027-06-30; the Stream v2 quote completed; 70-day restatement window and the 2026-09-30 general availability added.
+29. Practitioners, Ad Badger FTC post: "read through a search-result summary only" → fetched; exact title, 2026-09-10 date and quotes added.
+30. Trellis: example "$30 product, 30% conversion, $0.60 base bid and 10% target ACOS" → 30% target ACOS and 10% top-of-search conversion, with its other worked cases; "a ceiling, not a target" added.
+31. Suggested-bid vendors: the 25th–75th percentile reading attributed to Amalyze (fetched); Seller Metrics kept as search-only.
+32. BidX: placement comparisons are within down-only campaigns; ACOS wording "almost 5 percentage points"; BidX's own recommendations added.
+33. Perpetua (dynamic bidding): "since sales history doesn't transfer" was not verbatim → the page's own sentences, attributed as it relays "Amazon’s word of caution".
+34. Novadata: author role (CTO, Nova Analytics), title, quote and listed sources corrected; a single-source "Smart Bid V2" report noted as unverified.
+35. Perpetua (top of search): "can cannibalize their organic sales" was not verbatim → its ACOS/TACOS sentence quoted.
+36. Ad Badger (creator placements): option labels on Amazon's page given; Ad Badger's "Only your base bid governs delivery" set against Amazon's "Dynamic bidding and rule-based bidding still apply".
+37. Ad Badger (dayparting): modified 2026-08-12 added; its statements quoted.
+38. Research, Chen et al.: "Amazon Ads; Siniscalchi also Northwestern" → Chen at Outreach Inc., Nabi at Amazon Ads, Siniscalchi at Amazon Ads and Northwestern; the seat note now records the pricing difference above the floor.
+39. Research, Balseiro et al.: "reserves and boosts set from machine-learned signals" → the abstract's "appropriately chosen" reserves and additive boosts.
+40. Research citations: Agarwal et al. DOI and Blake et al. pages (155–174) added.
+41. Page names corrected throughout: *Scheduled bid rules*, *Theme-based bid suggestions - Quick-start guide*, *Reporting columns (version 3)*, *Search (Impression share)*, *Sponsored Products performance datasets*, *Getting started with campaigns*, *Amazon Ads API for Sponsored Display overview*, *Managing Sponsored Brands optimization rules*, *The campaign entity in the Amazon Ads API*, *Ads performance*, *Overview* (Stream v2), *Sponsored Products Rule-based bidding overview*.
+42. Pages read: names and contents corrected; the Deprecations page, Amalyze and the arXiv/Crossref records added; the FTC and Ad Badger rows annotated.
+43. Positions rewritten to match: P-8 (v1 model added), P-23 (impact metrics as shut-off v3 content), P-24 (v5 scope), P-29 (dates), P-31 (Sponsored Brands vs display cost control), P-35 (the $3 warning is the display help page's), P-36 (Stream v1 end and v2 window), P-37 (within down-only), P-40 (affiliations).
+44. Run-time reads: v5 estimates instead of impact metrics, the Stream v1 end date and Stream v2 as an hourly source, and v1's per-campaign time zone.
+45. Open questions 7 and 11 updated; 4 and 10 resolved.
+
+Removed:
+
+1. "Scale Insights likewise" (C): its page does not state a 100%/50% ceiling.
+2. The Trellis bullet on report placement values and "sunsetting Sponsored Ads reports ... by December 31, 2026": not on the page as updated 2026-11-04 (Amazon's Deprecations page gives 2027-06-30 for v3 and Sponsored Ads v2 reporting).
+
+Still unreachable after verification: the help node ids listed under *Unreachable* (including GPDRGYSKVB7H4BUR); the cloudfront markdown paths that return `NoSuchKey`; the Amazon Ads library guides that return Amazon Ads' 404 page; the aboutamazon `/news/` addresses and the Pinterest/BuzzFeed post; Pacvue (403); m19 (404); Seller Metrics (no readable text); Pulse (fetch failed); the FTC complaint itself; the AuctionGym PDF (not extracted); journal full texts; the Ad Lab video; and the MCP server's `/skills` download, not fetched because it sits on an Ads API host.
+
+The verifier called no Ads API host, no Login with Amazon endpoint, no Amazon Ads MCP server endpoint and no SP-API host.
+
+### Gap fill 2026-11-09
+
+Done by the research seat that owns this file, on three completeness critics' items. Every page cited below was opened on 2026-11-09 by plain GET (Ads API guides as cloudfront markdown, OpenAPI models from their cloudfront addresses, help articles from the JSON embedded in the page, news and aboutamazon pages as HTML) or, for two research pages, through a fetch tool; abstracts came from Crossref, arXiv and publisher pages. No Ads API host, Login with Amazon endpoint, Amazon Ads MCP endpoint, SP-API host or signed-in page was called, and no seller data was read.
+
+Added:
+
+1. Gap 1 (Alexa for Shopping). Section F retitled; three RULE bullets added: the 2026-05-13 merger (aboutamazon, two pages), the post-merger ad surfaces from *What agentic shopping means for advertising* (2026-06-11: automatic eligibility in Alexa for Shopping, prompts, Alexa+ Conversational Ads with automatic enrolment, "closed-loop measurement"), and what Amazon does not say: no page read states which bid, placement adjustment, audience boost or strategy applies there; the release notes have no Alexa for Shopping entry; the prompt reports carry no placement or bid column. The 2024 Rufus and 2025 Alexa+ notes are kept as dated history. Scope list, Stable 19, Open question 5, positions P-41 to P-43 and pages-read rows added.
+2. Gap 2 (bid writes). New section O with the write operations, body fields, media types, batch ceilings and 207 shapes for SP v3 (`PUT /sp/keywords`, `/sp/targets`, `/sp/adGroups`, `/sp/campaigns`), v1 (`POST /adsApi/v1/update/targets`, `/adGroups`, `/campaigns`, with per-product ceilings and enums), SB (`PUT /sb/keywords`, `/sb/targets`, `/sb/v4/campaigns`) and SD (`PUT /sd/targets`, `/sd/adGroups`), with the per-item refusal reasons the models give (`BID_OUT_OF_MARKET_PLACE_RANGE`, `BID_GT_BUDGET`, `BID_INVALID_PLACEMENT` and the audience reasons; `ARCHIVED_ENTITY_CANNOT_BE_MODIFIED`, `PARENT_ARCHIVED_FORBIDS_UPDATES`, `AUTO_TARGETING_CLAUSE_CANNOT_BE_ARCHIVED_MANUALLY`; v1 `ARCHIVED_RESOURCE_CANNOT_EDIT`, `ARCHIVED_PARENT_CANNOT_EDIT`, `AUTOCREATED_ENTITY_CANNOT_EDIT`, `FIELD_VALUE_IS_OUT_OF_RANGE`; the SP budget-below-bids reasons). The run-time section now names the writes and the read-back. Stable 17–18, Open questions 13–14, positions P-44 to P-48.
+   - Partly declined: the gap's wording that an SD bid write under cost control or an optimization rule "is accepted but has no effect". Amazon's models say the bid is "only mutable when the targeting clause's adGroup does not have any enabled optimization rule" and that manual target bids "will be overridden"; neither says whether the write is refused or accepted. The file records both wordings and leaves the outcome as Open question 13. The same optimization-rule sentence was found on the SB keyword and target writes and added.
+   - Discrepancies found in the models and recorded, not resolved: SD `PUT /sd/targets` states 100 and 1,000 as its ceiling; the per-product SP v1 model allows 1 audience adjustment (merged model: 10) and 4 placement adjustments (merged: 5); v1 `ErrorsIndex.index` is bounded 0–19; SB `PUT /sb/targets` answers 200, not 207.
+3. Gap 3 (position bias). New research subsections "Conversion rate by position: why the studies disagree" and "Position bias in click data and placement comparisons": Craswell, Zoeter, Taylor & Ramsey (WSDM 2008); Joachims, Swaminathan & Schnabel (WSDM 2017); Reiley, Li & Lewis (EC 2010); Rutz, Bucklin & Sonnier (JMR 2012); Karra et al. (CIKM 2023, amazon.science); Wang et al. (PACC). Positions P-51 to P-53 and a seat position on placement comparisons.
+   - Declined in part: the gap calls Xue et al. 2023 "Amazon's own position de-biasing". The paper's PDF gives the authors' affiliation as "Walmart eCommerce" (walmart.com addresses), so it is recorded as Walmart's system, not Amazon's; the follow-on PACC paper is by UIC, Walmart and Indeed authors. An Amazon-affiliated paper on position bias in click prediction (Karra et al., CIKM 2023) was added in its place. The note in `areas/campaign-structure.md` §6 that a vendor (Rel.ai) cites Xue et al. as describing Amazon's ranking rests on a paper that is Walmart's; that file's owner should know.
+   - Not confirmed: the gap calls Reiley, Li & Lewis "a Yahoo! experiment". Neither the SSRN abstract nor Lewis's MIT thesis abstract names the engine; the file says so.
+
+Settled:
+
+1. Contradiction 1 (Rufus and Alexa+ separate or merged). Merged: aboutamazon (2026-05-13, "Amazon brings together Rufus and Alexa+ to create ‘Alexa for Shopping’"; "On May 13, 2026, Rufus was renamed Alexa for Shopping") and Amazon Ads (2026-06-11) both say so. `beyond/ad-products-and-eligibility.md`'s observation that "Alexa for Shopping" appears on no ads page is overtaken by the Amazon Ads news page *What agentic shopping means for advertising*, which uses the name throughout; that file is the owner and should align. This file now treats the 2024 and 2025 notes as history (F).
+2. Contradiction 2 (reserve share of voice). Re-opened the three Amazon pages: 30–92 days and "not available in Spain" (*Campaign goals*), 15–92 days, $3,000 and "not available in Spain" (*Reserve keywords*), "e.g., 14–90 days", $6K US and 20 marketplaces including Ireland (library guide). K now carries the discrepancy and points to `beyond/brand-defense-and-competition.md`; P-33 rewritten.
+3. Contradiction 3 and 8 (up-and-down ceiling). This file is the owner. Re-opened: help GCU2BUWJH2W3A8Z7 ("up to 100% for all placements"), the library guide ("Currently, we can increase this bid by up to 100% … for all placements opportunities"), the SP 3.0 and v1 models ("by a maximum of 100%"), and the author's guide ("up to 100% for placements at the top of the first page … up to 50% for all other placements"); the *Set bids* help page gives no percentage. The fact is: four current Amazon sources say 100% for all placements and one Amazon page, written for KDP authors, says 100/50; no Amazon page dates a change. "The 50% is the older rule" is a reading (this seat's, PPC Land's), not Amazon's statement, and C now says so. P-8 stands as written. For other files: a single-value position ("up to 100% for all placements", `areas/launch.md` P-24, `areas/diagnosis.md` P-37, `beyond/automation-and-amazon-tools.md` P-2) is true of the four sources but should name the author's guide as the dissenting Amazon page or defer to P-bids-and-placements-8; `method/change-measurement.md` P-12 is accurate, and its Open question 8 can be closed as "Amazon's pages disagree four to one; the live value is not readable from any field", with no further page to read.
+4. Contradiction 4 (SD on v1). `method/ads-api-surface.md` is right on the facts: the campaign-management overview lists Sponsored Display as "currently" supported and the SD v1 model (re-read: update ceilings 100/1,000, `bid.defaultBid`, `bid.bid`, KPI enum) exists. L's "not yet in the unified v1 campaign API" was wrong and is replaced; what remains true is that no release note dates SD general availability and the entity guide's table has no SD column. P-49 added; Open question 15.
+5. Contradiction 5 (`topOfSearchImpressionShare` report types). Re-read the three v3 pages. The columns page lists `spCampaigns`, `sbCampaigns`, `sbTargeting`; the targeting report page lists it in SP and SB base metrics; the campaign report page lists it under SP "Group by `campaign`" and "Group by `campaignPlacement`" and in SB base metrics. This file's statement stands and is made exact (J, P-26). `method/reporting.md` §C's list (SP only) is incomplete against these pages; that file is the owner and should add `sbCampaigns` and `sbTargeting` and the `campaignPlacement` listing.
+6. Contradiction 6 (Stream v1 dataset id). `method/reporting.md` is right: the page whose path is `datasets/sp-performance` describes two datasets subscribed to as `sp-traffic` and `sp-conversion`. M and the run-time section corrected; P-50 added.
+7. Contradiction 7 (conversion rate by position). This file is the owner. Re-read the four abstracts through Crossref: Ghose & Yang (highest at the top), Rutz, Bucklin & Sonnier (higher positions raise both CTR and conversion), Agarwal, Hosanagar & Smith ("conversion rate increases with position", that is, lower down), Narayanan & Kalyanam (auction-set positions bias naive comparisons; causal effects smaller and only at some positions). Each file's single citation is accurate for its own paper; none is the literature's answer. The new subsection in *Published research* reconciles them, and P-51 states the disagreement. Other files citing one paper (`areas/search-terms-and-negatives.md` R5, `areas/keyword-and-product-targeting.md`, `beyond/brand-defense-and-competition.md`) can point here.
+
+Declined: none of the eight contradictions; gaps 2 and 3 in part, as stated above.