late-sdk 0.0.1365 → 0.0.1367

data/docs/AdCampaignsApi.md

@@ -76,7 +76,7 @@ All URIs are relative to *https://zernio.com/api*
 
 Add Search ad-group keywords
 
-Adds one or more keyword criteria to an existing Google Search ad group, without touching the keywords already there (unlike the whole-set diff on `PUT /v1/ads/{adId}`, `keywords`/`negativeKeywords` in `platformSpecificData`, which replaces the set). Set `negative: true` to add ad-group-level negatives instead of positive keywords. 
+Adds keywords, or negative keywords, to a Google Search ad group without replacing the ones already there.  Adds one or more keyword criteria to an existing Google Search ad group, without touching the keywords already there (unlike the whole-set diff on `PUT /v1/ads/{adId}`, `keywords`/`negativeKeywords` in `platformSpecificData`, which replaces the set). Set `negative: true` to add ad-group-level negatives instead of positive keywords. 
 
 ### Examples
 
@@ -145,7 +145,7 @@ end
 
 Apply Google Ads recommendations
 
-Apply up to 100 recommendations. This changes the account (budgets, bidding, keywords, assets) and is not reversible or idempotent; Google offers no validate-only mode for it. Items run in partial-failure mode, so one stale recommendation does not block the rest. `parameters` is optional and takes exactly one key named for the recommendation type, in Google's ApplyRecommendationOperation shape (for example `campaignBudget: { newBudgetAmountMicros }` or `keyword: { matchType, cpcBidMicros }`); omit it to apply Google's suggested values.
+Applies Google's optimization suggestions to an ad account, using Google's values or your own; the changes cannot be undone.  Apply up to 100 recommendations. This changes the account (budgets, bidding, keywords, assets) and is not reversible or idempotent; Google offers no validate-only mode for it. Items run in partial-failure mode, so one stale recommendation does not block the rest. `parameters` is optional and takes exactly one key named for the recommendation type, in Google's ApplyRecommendationOperation shape (for example `campaignBudget: { newBudgetAmountMicros }` or `keyword: { matchType, cpcBidMicros }`); omit it to apply Google's suggested values. 
 
 ### Examples
 
@@ -214,7 +214,7 @@ end
 
 Attach ad-group assets
 
-Creates and attaches sitelinks, callouts, structured snippets and image assets in one Google mutation. Google shows images only on accounts it deems eligible (account age, policy history, vertical).
+Creates sitelinks, callouts, structured snippets or image assets and attaches them to a Google ad group in one step.  Creates and attaches sitelinks, callouts, structured snippets and image assets in one Google mutation. Google shows images only on accounts it deems eligible (account age, policy history, vertical). 
 
 ### Examples
 
@@ -285,7 +285,7 @@ end
 
 Attach campaign assets
 
-Creates and attaches sitelinks, callouts, structured snippets and image assets in one Google mutation. Google shows images only on accounts it deems eligible (account age, policy history, vertical).
+Adds sitelinks, callouts, structured snippets and images to a Google campaign in one step.  Creates and attaches sitelinks, callouts, structured snippets and image assets in one Google mutation. Google shows images only on accounts it deems eligible (account age, policy history, vertical). 
 
 ### Examples
 
@@ -356,7 +356,7 @@ end
 
 Boost post as ad
 
-Creates a paid ad from an existing published post, keeping the post's engagement. By default it provisions the whole hierarchy (campaign, ad set, ad).  **Attach shape (Meta).** Send `adSetId` to put the ad under an EXISTING ad set instead, so that ad set keeps its learning phase. It then owns `budget`, `schedule` and `targeting`, and sending any of those alongside `adSetId` is a 400 rather than a silent drop. `budget` is required only without `adSetId`.  `instagramAccountId`, `destinationType`, `whatsappPhoneNumber` and `adSetId` are Meta-only and return 400 on other platforms.  `accountId` may be a Facebook, Instagram or Meta ads (business login) connection. A business-login connection has no posting account, so pass the post as `platformPostId` (Facebook `pageId_postId` or an Instagram media id); a Zernio `postId` is a 400 there.  For X, `accountId` may be the posting X account or its X Ads connection: either resolves the other on the same profile. A Zernio `postId` on an X Ads connection resolves through its X account, and is a 400 when the profile has none (pass `platformPostId` instead).  **Messaging boosts (Meta).** Use `goal: engagement` with `callToAction: WHATSAPP_MESSAGE`, `MESSAGE_PAGE`, or `INSTAGRAM_MESSAGE`. The CTA implies WHATSAPP, MESSENGER, or INSTAGRAM_DIRECT respectively; `destinationType` alone does not select a messaging CTA. Omit `linkUrl` only for messaging CTAs. Plain link CTAs keep their goal and link behavior when combined with an independent `destinationType`. The campaign uses OUTCOME_ENGAGEMENT and the ad set uses CONVERSATIONS with the promoted Page. Optional `whatsappPhoneNumber` selects a number already paired with that Page. Conflicting CTA/destination, instant form, goal, or optimizationGoal inputs return 400. Attach requires the target ad set destination to match. Existing post references preserve social proof; an Instagram reel rejected by Meta is not re-uploaded as a new post for a messaging boost.  **Retries.** Boosts are NOT idempotent and can take minutes when Meta requires re-hosting an Instagram video, so do not retry on client timeout. Send an Idempotency-Key header to make retries safe: same key and body replays the original 201, and distinct keys always create distinct ads. Without the header, an identical request is treated as a retry: while one is in flight it returns 409, and within 10 minutes of a completed boost it returns the already-created ad instead of creating another. To intentionally duplicate an ad, send distinct Idempotency-Keys (or vary the body, e.g. the name). 
+Turns an already published post into a paid ad that keeps the post's likes and comments, creating the campaign around it.  By default it provisions the whole hierarchy (campaign, ad set, ad).  **Attach shape (Meta).** Send `adSetId` to put the ad under an EXISTING ad set instead, so that ad set keeps its learning phase. It then owns `budget`, `schedule` and `targeting`, and sending any of those alongside `adSetId` is a 400 rather than a silent drop. `budget` is required only without `adSetId`.  `instagramAccountId`, `destinationType`, `whatsappPhoneNumber` and `adSetId` are Meta-only and return 400 on other platforms.  `accountId` may be a Facebook, Instagram or Meta ads (business login) connection. A business-login connection has no posting account, so pass the post as `platformPostId` (Facebook `pageId_postId` or an Instagram media id); a Zernio `postId` is a 400 there.  For X, `accountId` may be the posting X account or its X Ads connection: either resolves the other on the same profile. A Zernio `postId` on an X Ads connection resolves through its X account, and is a 400 when the profile has none (pass `platformPostId` instead).  **Messaging boosts (Meta).** Use `goal: engagement` with `callToAction: WHATSAPP_MESSAGE`, `MESSAGE_PAGE`, or `INSTAGRAM_MESSAGE`. The CTA implies WHATSAPP, MESSENGER, or INSTAGRAM_DIRECT respectively; `destinationType` alone does not select a messaging CTA. Omit `linkUrl` only for messaging CTAs. Plain link CTAs keep their goal and link behavior when combined with an independent `destinationType`. The campaign uses OUTCOME_ENGAGEMENT and the ad set uses CONVERSATIONS with the promoted Page. Optional `whatsappPhoneNumber` selects a number already paired with that Page. Conflicting CTA/destination, instant form, goal, or optimizationGoal inputs return 400. Attach requires the target ad set destination to match. Existing post references preserve social proof; an Instagram reel rejected by Meta is not re-uploaded as a new post for a messaging boost.  **Retries.** Boosts are NOT idempotent and can take minutes when Meta requires re-hosting an Instagram video, so do not retry on client timeout. Send an Idempotency-Key header to make retries safe: same key and body replays the original 201, and distinct keys always create distinct ads. Without the header, an identical request is treated as a retry: while one is in flight it returns 409, and within 10 minutes of a completed boost it returns the already-created ad instead of creating another. To intentionally duplicate an ad, send distinct Idempotency-Keys (or vary the body, e.g. the name). 
 
 ### Examples
 
@@ -429,7 +429,7 @@ end
 
 Pause or resume many campaigns
 
-Process up to 50 campaigns in one call. Each campaign is updated concurrently and the response contains a per-campaign result so a single bad row does not fail the whole batch. Each campaign is read, written and re-read exactly as PUT /v1/ads/campaigns/{campaignId}/status describes: only the campaign's own switch is written, never its ad sets' or ads'. `updated` / `skipped` count campaigns. 
+Pauses or resumes up to 50 campaigns in one call, with a separate result per campaign so one failure does not block the rest.  Each campaign is updated concurrently and the response contains a per-campaign result so a single bad row does not fail the whole batch. Each campaign is read, written and re-read exactly as PUT /v1/ads/campaigns/{campaignId}/status describes: only the campaign's own switch is written, never its ad sets' or ads'. `updated` / `skipped` count campaigns. 
 
 ### Examples
 
@@ -498,7 +498,7 @@ end
 
 Create a standalone campaign
 
-Creates a campaign WITHOUT its first ad set / ad, on the platform of the given `accountId`. Ad sets join it later via `existingCampaignId` on the create endpoints. Platform notes: on Meta a budget here is campaign-level (CBO) by definition; omit it for ABO (each ad set carries its own budget), and `specialAdCategories` is Meta-only (400 elsewhere); `bidStrategy` is Meta and Google (400 elsewhere), and Google also accepts `portfolioBidStrategyId` instead. Google, X and OpenAI require a budget (422 without one; OpenAI accepts daily or lifetime, Google only `budgetType: daily`). On OpenAI `goal` sets the campaign objective, and `conversions` needs an active standard conversion event on the account. LinkedIn creates the campaign GROUP (our campaign level) and rejects a budget, which lives on the campaign (ad set) level there; it comes back `status: DRAFT`. Created `PAUSED` (TikTok `DISABLE`) unless `status: ACTIVE` where the platform supports it.  **Idempotency:** send an `Idempotency-Key` header to make retries safe.
+Creates an empty campaign, paused, so you can add ad sets and ads to it afterwards.  Creates a campaign WITHOUT its first ad set / ad, on the platform of the given `accountId`. Ad sets join it later via `existingCampaignId` on the create endpoints. Platform notes: on Meta a budget here is campaign-level (CBO) by definition; omit it for ABO (each ad set carries its own budget), and `specialAdCategories` is Meta-only (400 elsewhere); `bidStrategy` is Meta and Google (400 elsewhere), and Google also accepts `portfolioBidStrategyId` instead. Google, X and OpenAI require a budget (422 without one; OpenAI accepts daily or lifetime, Google only `budgetType: daily`). On OpenAI `goal` sets the campaign objective, and `conversions` needs an active standard conversion event on the account. LinkedIn creates the campaign GROUP (our campaign level) and rejects a budget, which lives on the campaign (ad set) level there; it comes back `status: DRAFT`. Created `PAUSED` (TikTok `DISABLE`) unless `status: ACTIVE` where the platform supports it.  **Idempotency:** send an `Idempotency-Key` header to make retries safe.
 
 ### Examples
 
@@ -571,7 +571,7 @@ end
 
 Create a standalone ad group
 
-Google Ads compliance row C.190: creates an ad group WITHOUT an ad, under an existing campaign. Ads join it later via `adSetId` on POST /v1/ads/create. Google only; every other platform returns 501.  Created `PAUSED` unless `status: ACTIVE`. The new ad group has no ad yet, so it will not appear in GET /v1/ads/tree (built purely from `ads` rows) until one is added; use GET /v1/ads/ad-sets to see it in the meantime.  **Idempotency:** send an `Idempotency-Key` header to make retries safe.
+Creates an empty Google ad group under an existing campaign, ready for ads to be added later.  Google Ads compliance row C.190: creates an ad group WITHOUT an ad, under an existing campaign. Ads join it later via `adSetId` on POST /v1/ads/create. Google only; every other platform returns 501.  Created `PAUSED` unless `status: ACTIVE`. The new ad group has no ad yet, so it will not appear in GET /v1/ads/tree (built purely from `ads` rows) until one is added; use GET /v1/ads/ad-sets to see it in the meantime.  **Idempotency:** send an `Idempotency-Key` header to make retries safe.
 
 ### Examples
 
@@ -644,7 +644,7 @@ end
 
 Create portfolio bid strategy
 
-Creates a standalone bid strategy shared across campaigns. Attach it to a campaign with `portfolioBidStrategyId` on POST /v1/ads/create, PUT /v1/ads/campaigns/{campaignId}, or PUT /v1/ads/ad-sets/{adSetId}. Attaching a strategy aligned to a shared budget fails there with a 400 (Google's `BIDDING_STRATEGY_AND_BUDGET_MUST_BE_ALIGNED`); this is not retryable.
+Creates a Google Ads bid strategy that several campaigns can share, such as one target CPA or ROAS.  Attach it to a campaign with `portfolioBidStrategyId` on POST /v1/ads/create, PUT /v1/ads/campaigns/{campaignId}, or PUT /v1/ads/ad-sets/{adSetId}. Attaching a strategy aligned to a shared budget fails there with a 400 (Google's `BIDDING_STRATEGY_AND_BUDGET_MUST_BE_ALIGNED`); this is not retryable. 
 
 ### Examples
 
@@ -713,7 +713,7 @@ end
 
 Create a Performance Max asset group
 
-Add an asset group to an existing Performance Max campaign. The group, any new assets, their links and an optional listing-group tree are created in one atomic request, so Google checks the asset minimums (for non-retail campaigns) against the whole set. Created PAUSED unless status is ENABLED. validateOnly: true runs Google's validation without creating anything.
+Adds an asset group with its assets and optional product targeting to a Performance Max campaign, or just validates it without creating anything.  The group, any new assets, their links and an optional listing-group tree are created in one atomic request, so Google checks the asset minimums (for non-retail campaigns) against the whole set. Created PAUSED unless status is ENABLED. validateOnly: true runs Google's validation without creating anything. 
 
 ### Examples
 
@@ -784,7 +784,7 @@ end
 
 Create a shared budget
 
-Creates a daily shared campaign budget (`explicitly_shared: true`, standard delivery) that several campaigns can draw from. A lifetime budget returns 422, like every Google budget. Google refuses some bidding strategies on a shared budget; that error surfaces when a campaign is moved onto it. 
+Creates a daily Google Ads budget that several campaigns can share.  Creates a daily shared campaign budget (`explicitly_shared: true`, standard delivery) that several campaigns can draw from. A lifetime budget returns 422, like every Google budget. Google refuses some bidding strategies on a shared budget; that error surfaces when a campaign is moved onto it. 
 
 ### Examples
 
@@ -853,7 +853,7 @@ end
 
 Create standalone ad
 
-Create a paid ad with custom creative across Meta, Google Ads, Pinterest, TikTok, X, LinkedIn, and OpenAI Ads (ChatGPT Ads).  Google Performance Max: set `campaignType: \"pmax\"` and supply `assetGroup` with text, images by role, business name and finalUrl. Creates a daily budget, PAUSED campaign and asset group atomically. `validateOnly: true` validates the complete request with Google without creating or persisting resources. Read assets with `GET /v1/ads/campaigns/{campaignId}/asset-groups`. The logo is required; video is optional via `assetGroup.youtubeVideoId`. Brand guidelines are disabled at creation. All supplied asset links are validated together against Google's minimum asset requirements. PMax rejects ACTIVE creation, portfolio bidding, bid caps, legacy creative fields and attach shapes. Geo and language targeting are supported; omitted geo targets all locations. PMax does not require top-level goal, headline, body or linkUrl. Supported bidding: omitted or LOWEST_COST_WITHOUT_CAP for Maximize Conversions, COST_CAP plus bidAmount for target CPA, LOWEST_COST_WITH_MIN_ROAS plus roasAverageFloor for Maximize Conversion Value with target ROAS.  Google Demand Gen: set `campaignType: \"demand_gen\"` and supply `demandGen`. Creates a daily budget, PAUSED campaign, one ad group and one ad in a single atomic request: a multi-asset image ad, a video responsive ad when `demandGen.youtubeVideoIds` is sent, or a carousel ad (2 to 10 cards) when `demandGen.carouselCards` is sent. Geo (countries, regions, cities, zips, metros) and languages go on the ad group, as Demand Gen requires. `demandGen.channels` sets the ad group's channel controls and `demandGen.audience` builds a Google Audience (user lists, interests, custom audiences, age ranges, genders) attached to the ad group, or `demandGen.audienceId` attaches an existing one. `validateOnly: true` validates everything with Google without creating resources. Supported bidding: omitted or LOWEST_COST_WITHOUT_CAP for Maximize Conversions, COST_CAP plus bidAmount for target CPA, LOWEST_COST_WITH_MIN_ROAS plus roasAverageFloor for target ROAS, LOWEST_COST_WITH_BID_CAP plus bidAmount for target CPC. The created ad carries the native `platformCampaignId`, `platformAdSetId` (ad group) and `platformAdId`. Activate with PUT /v1/ads/campaigns/{campaignId}/status; budget, bidding and name are edited with the regular campaign endpoints; creative, channels, audience and ad group targeting with PUT /v1/ads/{adId} (`demandGen`, `targeting`). To grow an existing Demand Gen campaign, send `existingCampaignId` (adds a PAUSED ad group, with its own geo, languages, channels and audience, plus its ad) or `adSetId` (adds a PAUSED ad to that ad group); neither takes budget, bidding or schedule fields, and both support `validateOnly`.  Other mutually-exclusive request shapes are selected by the body:  - Legacy single-creative shape (all platforms, the default). - Meta-only multi-creative shape via the creatives array: one ad set with N ads sharing budget and targeting. - Attach shape via adSetId: adds one new ad to an existing ad set, inheriting its budget, targeting, and schedule (Meta, Google Ads, TikTok, and LinkedIn). On LinkedIn adSetId is the existing Campaign id, and the budget, schedule, targeting and bidding fields must be omitted. On LinkedIn `goal` may be omitted too (taken from the Campaign's objective), and the created ad's `targeting` echoes the Campaign's audience.  Meta accepts `creativeFeatures` on the single and attach shapes and as defaults for `creatives[]`; an item replaces the whole feature map. `promotion` is not supported on any shape and any object is rejected with 400. Reusing `existingCreativeId` uses the existing creative settings instead of new settings. Requested settings are persisted for lists, exports, and default ad-detail reads.  Per-platform required fields, budget minimums, and video-ad rules are documented on each property below.  LinkedIn creates a Single Image or Single Video Ad backed by a Direct Sponsored Content \"dark post\" authored by a Company Page (see `organizationId`). Supported goals are engagement, traffic, awareness, and video_views (video ads use the `video` field; video_views requires a video), and traffic ads require `linkUrl`.  **Idempotency:** this endpoint is not idempotent at the platform level (a blind retry creates a second campaign/ad set/ad). Send an `Idempotency-Key` header to make retries safe: the first request with a given key creates the ad and we store the response; a retry with the same key replays that exact response (with `Idempotent-Replayed: true`) instead of creating duplicates. Reusing a key with a different body returns 422; a key whose first request is still in flight returns 409 (retry after a short backoff). Keys are scoped to your credential and expire after 24h. 
+Creates a paid ad with your own creative, building the campaign and ad set it needs or adding it to existing ones.  Create a paid ad with custom creative across Meta, Google Ads, Pinterest, TikTok, X, LinkedIn, and OpenAI Ads (ChatGPT Ads).  Google Performance Max: set `campaignType: \"pmax\"` and supply `assetGroup` with text, images by role, business name and finalUrl. Creates a daily budget, PAUSED campaign and asset group atomically. `validateOnly: true` validates the complete request with Google without creating or persisting resources. Read assets with `GET /v1/ads/campaigns/{campaignId}/asset-groups`. The logo is required; video is optional via `assetGroup.youtubeVideoId`. Brand guidelines are disabled at creation. All supplied asset links are validated together against Google's minimum asset requirements. PMax rejects ACTIVE creation, portfolio bidding, bid caps, legacy creative fields and attach shapes. Geo and language targeting are supported; omitted geo targets all locations. PMax does not require top-level goal, headline, body or linkUrl. Supported bidding: omitted or LOWEST_COST_WITHOUT_CAP for Maximize Conversions, COST_CAP plus bidAmount for target CPA, LOWEST_COST_WITH_MIN_ROAS plus roasAverageFloor for Maximize Conversion Value with target ROAS.  Google Demand Gen: set `campaignType: \"demand_gen\"` and supply `demandGen`. Creates a daily budget, PAUSED campaign, one ad group and one ad in a single atomic request: a multi-asset image ad, a video responsive ad when `demandGen.youtubeVideoIds` is sent, or a carousel ad (2 to 10 cards) when `demandGen.carouselCards` is sent. Geo (countries, regions, cities, zips, metros) and languages go on the ad group, as Demand Gen requires. `demandGen.channels` sets the ad group's channel controls and `demandGen.audience` builds a Google Audience (user lists, interests, custom audiences, age ranges, genders) attached to the ad group, or `demandGen.audienceId` attaches an existing one. `validateOnly: true` validates everything with Google without creating resources. Supported bidding: omitted or LOWEST_COST_WITHOUT_CAP for Maximize Conversions, COST_CAP plus bidAmount for target CPA, LOWEST_COST_WITH_MIN_ROAS plus roasAverageFloor for target ROAS, LOWEST_COST_WITH_BID_CAP plus bidAmount for target CPC. The created ad carries the native `platformCampaignId`, `platformAdSetId` (ad group) and `platformAdId`. Activate with PUT /v1/ads/campaigns/{campaignId}/status; budget, bidding and name are edited with the regular campaign endpoints; creative, channels, audience and ad group targeting with PUT /v1/ads/{adId} (`demandGen`, `targeting`). To grow an existing Demand Gen campaign, send `existingCampaignId` (adds a PAUSED ad group, with its own geo, languages, channels and audience, plus its ad) or `adSetId` (adds a PAUSED ad to that ad group); neither takes budget, bidding or schedule fields, and both support `validateOnly`.  Other mutually-exclusive request shapes are selected by the body:  - Legacy single-creative shape (all platforms, the default). - Meta-only multi-creative shape via the creatives array: one ad set with N ads sharing budget and targeting. - Attach shape via adSetId: adds one new ad to an existing ad set, inheriting its budget, targeting, and schedule (Meta, Google Ads, TikTok, and LinkedIn). On LinkedIn adSetId is the existing Campaign id, and the budget, schedule, targeting and bidding fields must be omitted. On LinkedIn `goal` may be omitted too (taken from the Campaign's objective), and the created ad's `targeting` echoes the Campaign's audience.  Meta accepts `creativeFeatures` on the single and attach shapes and as defaults for `creatives[]`; an item replaces the whole feature map. `promotion` is not supported on any shape and any object is rejected with 400. Reusing `existingCreativeId` uses the existing creative settings instead of new settings. Requested settings are persisted for lists, exports, and default ad-detail reads.  Per-platform required fields, budget minimums, and video-ad rules are documented on each property below.  LinkedIn creates a Single Image or Single Video Ad backed by a Direct Sponsored Content \"dark post\" authored by a Company Page (see `organizationId`). Supported goals are engagement, traffic, awareness, and video_views (video ads use the `video` field; video_views requires a video), and traffic ads require `linkUrl`.  **Idempotency:** this endpoint is not idempotent at the platform level (a blind retry creates a second campaign/ad set/ad). Send an `Idempotency-Key` header to make retries safe: the first request with a given key creates the ad and we store the response; a retry with the same key replays that exact response (with `Idempotent-Replayed: true`) instead of creating duplicates. Reusing a key with a different body returns 422; a key whose first request is still in flight returns 409 (retry after a short backoff). Keys are scoped to your credential and expire after 24h. 
 
 ### Examples
 
@@ -926,7 +926,7 @@ end
 
 Cancel an ad
 
-Deletes the ad on the platform and marks it as cancelled in the database. The ad is preserved for history. OpenAI Ads has no delete API; the ad is archived instead (a terminal state, the closest equivalent).  Only the ad is deleted; a campaign or ad set is never deleted while it still holds another ad. On Meta, when the ad set and campaign were created by Zernio (by `POST /v1/ads/create` or `POST /v1/ads/boost`) and Meta lists no other ad in the ad set (archived ads count), the emptied ad set is deleted too, then the campaign once it holds no other ad set. Parents created outside Zernio, and the parents of an ad imported from the platform, are always kept. On LinkedIn the ad is a creative inside a campaign inside a campaign group: the creative is always removed, the campaign only when Zernio created it and LinkedIn lists no other creative in it that is not archived, canceled or deleted, and the campaign group only when Zernio created it and LinkedIn lists no other campaign in it that is not canceled or deleted (archived campaigns count). If the creative itself cannot be removed the request fails and no parent is touched. To delete a whole campaign on purpose use `DELETE /v1/ads/campaigns/{campaignId}`. 
+Deletes an ad on the platform while keeping its history, and removes empty parents only when they were created through this API.  Deletes the ad on the platform and marks it as cancelled in the database. The ad is preserved for history. OpenAI Ads has no delete API; the ad is archived instead (a terminal state, the closest equivalent).  Only the ad is deleted; a campaign or ad set is never deleted while it still holds another ad. On Meta, when the ad set and campaign were created by Zernio (by `POST /v1/ads/create` or `POST /v1/ads/boost`) and Meta lists no other ad in the ad set (archived ads count), the emptied ad set is deleted too, then the campaign once it holds no other ad set. Parents created outside Zernio, and the parents of an ad imported from the platform, are always kept. On LinkedIn the ad is a creative inside a campaign inside a campaign group: the creative is always removed, the campaign only when Zernio created it and LinkedIn lists no other creative in it that is not archived, canceled or deleted, and the campaign group only when Zernio created it and LinkedIn lists no other campaign in it that is not canceled or deleted (archived campaigns count). If the creative itself cannot be removed the request fails and no parent is touched. To delete a whole campaign on purpose use `DELETE /v1/ads/campaigns/{campaignId}`. 
 
 ### Examples
 
@@ -995,7 +995,7 @@ end
 
 Delete a campaign
 
-Deletes the whole campaign on the platform, cascading to its ad sets and ads. Locally, all Ad documents for this campaign are marked `status: cancelled`.  **Empty campaigns.** A campaign with zero ads has no local Ad documents to resolve, so it is invisible to `/v1/ads/tree` and this endpoint would 404. That state is produced by the two-step create flow (campaign, then ads via `existingCampaignId`) whenever Meta rejects the ad step. To delete such a shell, send `accountId` in the body: we skip the local lookup entirely and forward the delete to Meta. `accountId` is ignored when the campaign does have ads. 
+Deletes a campaign along with all of its ad sets and ads, including a campaign that never got any ads.  Locally, all Ad documents for this campaign are marked `status: cancelled`.  **Empty campaigns.** A campaign with zero ads has no local Ad documents to resolve, so it is invisible to `/v1/ads/tree` and this endpoint would 404. That state is produced by the two-step create flow (campaign, then ads via `existingCampaignId`) whenever Meta rejects the ad step. To delete such a shell, send `accountId` in the body: we skip the local lookup entirely and forward the delete to Meta. `accountId` is ignored when the campaign does have ads. 
 
 ### Examples
 
@@ -1066,7 +1066,7 @@ end
 
 Delete an ad set
 
-Deletes the ad set on the platform, cascading to its ads only (never the campaign). Locally, every Ad document under the ad set is marked `status: cancelled`.  Delete is soft on platforms that have no hard delete: LinkedIn moves the campaign to `PENDING_DELETION`, Pinterest archives the ad group, and X soft-flags the line item. Google removes the ad group. All remain readable for reporting. 
+Deletes an ad set and its ads while keeping the campaign, archiving it instead on platforms without a hard delete.  Locally, every Ad document under the ad set is marked `status: cancelled`.  Delete is soft on platforms that have no hard delete: LinkedIn moves the campaign to `PENDING_DELETION`, Pinterest archives the ad group, and X soft-flags the line item. Google removes the ad group. All remain readable for reporting. 
 
 ### Examples
 
@@ -1135,7 +1135,7 @@ end
 
 Dismiss Google Ads recommendations
 
-Dismiss up to 100 recommendations so Google stops suggesting them. Items run in partial-failure mode.
+Dismisses Google optimization suggestions so Google stops showing them.  Dismiss up to 100 recommendations so Google stops suggesting them. Items run in partial-failure mode. 
 
 ### Examples
 
@@ -1204,7 +1204,7 @@ end
 
 Duplicate an ad
 
-Duplicates a single ad via Meta's native `POST /{ad-id}/copies`. The copy is created paused. `adSetId` retargets the copy into another ad set; omitted = the source's own ad set. Accepts the Zernio ad id or the platform ad id. Sync discovery is triggered automatically (`syncAfter: false` to skip). Creative settings returned by Meta, including explicit promotion metadata and creativeFeatures, are preserved when the native copy requires a creative rebuild. Metadata Meta does not return cannot be recovered. When Meta refuses the native copy with its capability error (code 3), which happens for some creatives built by other tools, the ad is rebuilt instead: a new creative from the source's returned spec and a new ad in the target ad set, carrying the source name, status option, rename options and tracking specs.
+Copies a single Meta ad, optionally into another ad set, created paused so you can review it before launch.  Duplicates a single ad via Meta's native `POST /{ad-id}/copies`. The copy is created paused. `adSetId` retargets the copy into another ad set; omitted = the source's own ad set. Accepts the Zernio ad id or the platform ad id. Sync discovery is triggered automatically (`syncAfter: false` to skip). Creative settings returned by Meta, including explicit promotion metadata and creativeFeatures, are preserved when the native copy requires a creative rebuild. Metadata Meta does not return cannot be recovered. When Meta refuses the native copy with its capability error (code 3), which happens for some creatives built by other tools, the ad is rebuilt instead: a new creative from the source's returned spec and a new ad in the target ad set, carrying the source name, status option, rename options and tracking specs.
 
 ### Examples
 
@@ -1279,7 +1279,7 @@ end
 
 Duplicate a campaign
 
-Duplicates a campaign, including its ad sets, ads, creatives, and targeting by default (`deepCopy: true`). The copy is created paused so callers can review before launching.  Per-platform implementation: - **Meta** uses the native `POST /{campaign-id}/copies` endpoint. - **TikTok** has no native copy primitive; Zernio walks the source   graph (`/v2/campaign/get/`, `/v2/adgroup/get/`, `/v2/ad/get/`) and   recreates each entity via the corresponding `/create/` endpoints,   carrying over budget / targeting / bid_type / bid_price /   deep_bid_type / creative fields. Spark Ad linkage (`tiktok_item_id`)   is preserved. - **LinkedIn** has no native copy primitive; Zernio walks the source   CampaignGroup → Campaigns → Creatives and recreates each entity,   carrying over `type` / `costType` / `unitCost` /   `optimizationTargetType` / `creativeSelection` / `objectiveType` /   `format` / `dailyBudget` / `totalBudget` / `targetingCriteria` /   `runSchedule` and every Creative's `content` object verbatim.   `statusOption: INHERITED_FROM_SOURCE` is evaluated **per entity**:   any Group / Campaign / Creative whose source is `ACTIVE` gets its   clone activated too. Duplicating an ACTIVE campaign with   `INHERITED_FROM_SOURCE` starts a second front of spend the moment   the clone activates. The safe default is `PAUSED`.  The new hierarchy is asynchronous to materialize in our DB, and we trigger sync discovery automatically. Set `syncAfter: false` to skip and poll `/v1/ads/tree` on your own cadence.  Other platforms return 501 Not Implemented. 
+Copies a campaign with its ad sets, ads, creatives and targeting, created paused so you can review it before launch.  Duplicates a campaign, including its ad sets, ads, creatives, and targeting by default (`deepCopy: true`). The copy is created paused so callers can review before launching.  Per-platform implementation: - **Meta** uses the native `POST /{campaign-id}/copies` endpoint. - **TikTok** has no native copy primitive; Zernio walks the source   graph (`/v2/campaign/get/`, `/v2/adgroup/get/`, `/v2/ad/get/`) and   recreates each entity via the corresponding `/create/` endpoints,   carrying over budget / targeting / bid_type / bid_price /   deep_bid_type / creative fields. Spark Ad linkage (`tiktok_item_id`)   is preserved. - **LinkedIn** has no native copy primitive; Zernio walks the source   CampaignGroup → Campaigns → Creatives and recreates each entity,   carrying over `type` / `costType` / `unitCost` /   `optimizationTargetType` / `creativeSelection` / `objectiveType` /   `format` / `dailyBudget` / `totalBudget` / `targetingCriteria` /   `runSchedule` and every Creative's `content` object verbatim.   `statusOption: INHERITED_FROM_SOURCE` is evaluated **per entity**:   any Group / Campaign / Creative whose source is `ACTIVE` gets its   clone activated too. Duplicating an ACTIVE campaign with   `INHERITED_FROM_SOURCE` starts a second front of spend the moment   the clone activates. The safe default is `PAUSED`.  The new hierarchy is asynchronous to materialize in our DB, and we trigger sync discovery automatically. Set `syncAfter: false` to skip and poll `/v1/ads/tree` on your own cadence.  Other platforms return 501 Not Implemented. 
 
 ### Examples
 
@@ -1354,7 +1354,7 @@ end
 
 Duplicate an ad set
 
-Duplicates an ad set. The copy is created paused so callers can review before launching. `campaignId` retargets the copy into another campaign; omitted = the source's own campaign.  Meta: ads and creatives are included by default (`deepCopy: true`) via Meta's native `POST /{adset-id}/copies`; the new hierarchy materializes asynchronously and sync discovery is triggered automatically (`syncAfter: false` to skip).  TikTok: the ad group is read and recreated under the campaign with its targeting, bidding, budget and schedule (start reset to now); `deepCopy: true` recreates its ads too (default false). `startTime`, `endTime` and `renameStrategy` are ignored and `statusOption` must be PAUSED or absent. The copy appears on the next discovery sync.
+Copies an ad set, optionally with its ads and into another campaign, created paused so you can review it before launch.  The copy is created paused so callers can review before launching. `campaignId` retargets the copy into another campaign; omitted = the source's own campaign.  Meta: ads and creatives are included by default (`deepCopy: true`) via Meta's native `POST /{adset-id}/copies`; the new hierarchy materializes asynchronously and sync discovery is triggered automatically (`syncAfter: false` to skip).  TikTok: the ad group is read and recreated under the campaign with its targeting, bidding, budget and schedule (start reset to now); `deepCopy: true` recreates its ads too (default false). `startTime`, `endTime` and `renameStrategy` are ignored and `statusOption` must be PAUSED or absent. The copy appears on the next discovery sync. 
 
 ### Examples
 
@@ -1429,7 +1429,7 @@ end
 
 Link or unlink asset group assets
 
-Link existing assets or new content to the asset group, and unlink assets, in one atomic request. Links are applied before unlinks, so swapping the last asset of a role does not trip Google's per-role minimum. Unlinking removes the link only; the asset stays in the account library. validateOnly: true validates without writing.
+Links new or existing assets to a Performance Max asset group and unlinks others in one change, keeping unlinked assets in the account library.  Link existing assets or new content to the asset group, and unlink assets, in one atomic request. Links are applied before unlinks, so swapping the last asset of a role does not trip Google's per-role minimum. Unlinking removes the link only; the asset stays in the account library. validateOnly: true validates without writing. 
 
 ### Examples
 
@@ -1502,7 +1502,7 @@ end
 
 Get ad details
 
-Returns an ad with its creative, targeting, status, and performance metrics. Google Search ads include current creative.headlines, creative.descriptions and creative.finalUrls, preserving pinnedField. Top-level cachedAt and stale report cache freshness. Google mutations invalidate this read. RSA enrichment requires a stored advertisingChannelType of SEARCH. Ads with an unknown or other channel return their stored details without a Google read. If RSA enrichment fails, the stored ad is returned with HTTP 200 and without cache metadata.  The `{adId}` path segment accepts any identifier dialect Zernio indexes for the ad: - the Zernio internal `_id` (24-char hex) - Meta's numeric `platformAdId` (the value shipped in `comment.received` webhooks as `comment.ad.id`) - the creative's `effective_object_story_id` (`{pageId}_{postId}` shape, Facebook side) - the creative's `effective_instagram_media_id` (Instagram side)  Any of the four resolve to the same ad. Caller doesn't need a translation step. `creative.creativeFeatures` holds the stored requested settings, which do not confirm platform application.  **Status freshness.** `status`, `configuredStatus`, `platformStatus`, `platformAdSetStatus` and `platformCampaignStatus` are the values Zernio last stored. Background sync refreshes them, typically within 15 to 60 minutes (Google up to about 3 hours), and ended or long-paused objects may be refreshed less often. Zernio's own status writes re-read the switches they change. A change made in the platform's own ads manager therefore shows up here only after the next sync. Controllers that act on a switch should pass `live=true`, which reads the switches from the platform now, stores them, and returns `statusReadAt` (null when the read failed and the stored values were returned). With `live=true` the ad's own switch (`configuredStatus`), its delivery status (`status`, `platformStatus`) and its ad set and campaign switches (`platformAdSetStatus`, `platformCampaignStatus`) are read live. 
+Shows one ad with its creative, targeting, status and performance metrics.  Google Search ads include current creative.headlines, creative.descriptions and creative.finalUrls, preserving pinnedField. Top-level cachedAt and stale report cache freshness. Google mutations invalidate this read. RSA enrichment requires a stored advertisingChannelType of SEARCH. Ads with an unknown or other channel return their stored details without a Google read. If RSA enrichment fails, the stored ad is returned with HTTP 200 and without cache metadata.  The `{adId}` path segment accepts any identifier dialect Zernio indexes for the ad: - the Zernio internal `_id` (24-char hex) - Meta's numeric `platformAdId` (the value shipped in `comment.received` webhooks as `comment.ad.id`) - the creative's `effective_object_story_id` (`{pageId}_{postId}` shape, Facebook side) - the creative's `effective_instagram_media_id` (Instagram side)  Any of the four resolve to the same ad. Caller doesn't need a translation step. `creative.creativeFeatures` holds the stored requested settings, which do not confirm platform application.  **Status freshness.** `status`, `configuredStatus`, `platformStatus`, `platformAdSetStatus` and `platformCampaignStatus` are the values Zernio last stored. Background sync refreshes them, typically within 15 to 60 minutes (Google up to about 3 hours), and ended or long-paused objects may be refreshed less often. Zernio's own status writes re-read the switches they change. A change made in the platform's own ads manager therefore shows up here only after the next sync. Controllers that act on a switch should pass `live=true`, which reads the switches from the platform now, stores them, and returns `statusReadAt` (null when the read failed and the stored values were returned). With `live=true` the ad's own switch (`configuredStatus`), its delivery status (`status`, `platformStatus`) and its ad set and campaign switches (`platformAdSetStatus`, `platformCampaignStatus`) are read live. 
 
 ### Examples
 
@@ -1575,7 +1575,7 @@ end
 
 Get live campaign details
 
-Reads one campaign live from Meta, returned verbatim, so a caller that knows a campaign id no longer has to page `GET /v1/ads/campaigns` to find it. The default projection covers name, status, objective, buying type, bid strategy, budgets, spend cap, schedule and `issues_info`. `fields` is a raw-passthrough override; unknown fields return Meta's 400 verbatim. A campaign the resolved connection cannot see comes back as Meta's own 400, not a 404.
+Shows a Meta campaign's current settings straight from Meta, such as its objective, budget, bid strategy, schedule and delivery issues.  Reads one campaign live from Meta, returned verbatim, so a caller that knows a campaign id no longer has to page `GET /v1/ads/campaigns` to find it. The default projection covers name, status, objective, buying type, bid strategy, budgets, spend cap, schedule and `issues_info`. `fields` is a raw-passthrough override; unknown fields return Meta's 400 verbatim. A campaign the resolved connection cannot see comes back as Meta's own 400, not a 404.
 
 ### Examples
 
@@ -1650,7 +1650,7 @@ end
 
 Read the platform's review verdict for an ad
 
-Reads the ad's review verdict from the platform now: whether it was approved, where it may not deliver, and every rejection reason with TikTok's suggestion and the piece of content it refers to. Read-only, so it works on a paused ad without re-enabling it.  **Google**: reads `ad_group_ad.policy_summary` live. `approvalStatus` and `reviewStatus` are Google's verbatim (`approvalStatus`: APPROVED, APPROVED_LIMITED, AREA_OF_INTEREST_ONLY, DISAPPROVED, UNKNOWN; `reviewStatus`: REVIEW_IN_PROGRESS, REVIEWED, UNDER_APPEAL, ELIGIBLE_MAY_SERVE); `approved` is true for the three approved statuses, false for DISAPPROVED, null otherwise. `policyTopics` carries every policy topic entry with its `type` (PROHIBITED, LIMITED, ...) and Google's `evidences` and `constraints` verbatim; each PROHIBITED topic is also listed in `rejections` (reason = the topic). The `forbidden*` arrays are TikTok-only and always empty on Google.  TikTok uses `/ad/review_info/`; every other platform returns 501. On TikTok, use it alongside the ad's `platformStatus`: TikTok reports `AD_STATUS_AUDIT` while the ad is in review and `AD_STATUS_AD_PRE_ONLINE` once it passed and is about to deliver (both map to `status: pending_review`); `AD_STATUS_AUDIT_DENY` maps to `rejected`.
+Shows whether the platform approved an ad, where it may not deliver, and every rejection reason, without re-enabling a paused ad.  Reads the ad's review verdict from the platform now: whether it was approved, where it may not deliver, and every rejection reason with TikTok's suggestion and the piece of content it refers to. Read-only, so it works on a paused ad without re-enabling it.  **Google**: reads `ad_group_ad.policy_summary` live. `approvalStatus` and `reviewStatus` are Google's verbatim (`approvalStatus`: APPROVED, APPROVED_LIMITED, AREA_OF_INTEREST_ONLY, DISAPPROVED, UNKNOWN; `reviewStatus`: REVIEW_IN_PROGRESS, REVIEWED, UNDER_APPEAL, ELIGIBLE_MAY_SERVE); `approved` is true for the three approved statuses, false for DISAPPROVED, null otherwise. `policyTopics` carries every policy topic entry with its `type` (PROHIBITED, LIMITED, ...) and Google's `evidences` and `constraints` verbatim; each PROHIBITED topic is also listed in `rejections` (reason = the topic). The `forbidden*` arrays are TikTok-only and always empty on Google.  **Meta** (Facebook and Instagram ads): reads the Ad's `ad_review_feedback`, `effective_status` and `issues_info` live. `reviewStatus` is Meta's `effective_status` verbatim (ACTIVE, PAUSED, PENDING_REVIEW, DISAPPROVED, PREAPPROVED, WITH_ISSUES, IN_PROCESS, ...); `approved` is true for ACTIVE and PREAPPROVED, false for DISAPPROVED, null otherwise. `reviewFeedback` carries Meta's feedback verbatim: `global` (reason -> explanation) and `placementSpecific` (placement -> reason -> explanation). Each global reason is listed in `rejections`, and so is each placement-specific reason, with that placement in its `forbiddenPlacements`; the review's `forbiddenPlacements` lists every placement with feedback. When Meta returns no feedback (usually an ad in review or approved everywhere), the status comes back with empty `rejections` and empty `reviewFeedback` maps; an ACTIVE ad rejected on one placement still carries that placement's feedback. A deleted ad answers 404. `issues` is Meta's `issues_info` verbatim. `forbiddenAges`, `forbiddenLocations` and `forbiddenOperatingSystems` are always empty on Meta.  TikTok uses `/ad/review_info/`. Platforms other than TikTok, Google and Meta return 501. On TikTok, use it alongside the ad's `platformStatus`: TikTok reports `AD_STATUS_AUDIT` while the ad is in review and `AD_STATUS_AD_PRE_ONLINE` once it passed and is about to deliver (both map to `status: pending_review`); `AD_STATUS_AUDIT_DENY` maps to `rejected`.
 
 ### Examples
 
@@ -1719,7 +1719,7 @@ end
 
 Get live ad-set details
 
-Reads the ad set live from Meta, returned verbatim. The default projection includes `learning_stage_info` (learning-phase status: LEARNING / SUCCESS / FAIL / WAIVING; Meta omits its `status` key on paused ad sets), delivery settings, budgets, schedule and targeting. `fields` is a raw-passthrough override; unknown fields return Meta's 400 verbatim.
+Shows a Meta ad set's current settings straight from Meta, including its learning phase, budget, schedule and targeting.  Reads the ad set live from Meta, returned verbatim. The default projection includes `learning_stage_info` (learning-phase status: LEARNING / SUCCESS / FAIL / WAIVING; Meta omits its `status` key on paused ad sets), delivery settings, budgets, schedule and targeting. `fields` is a raw-passthrough override; unknown fields return Meta's 400 verbatim.
 
 ### Examples
 
@@ -1794,7 +1794,7 @@ end
 
 Get campaign tree
 
-Returns a nested Campaign > Ad Set > Ad hierarchy with rolled-up metrics at each level. Uses a two-stage aggregation: ads are grouped into ad sets, then ad sets into campaigns. Metrics are computed over an optional date range, then rolled up from ad level to ad set and campaign levels. Pagination is at the campaign level. Ads without a campaign or ad set ID are grouped into synthetic \"Ungrouped\" buckets. If no date range is provided, defaults to the last 90 days. Date range is capped at 730 days max.  Pass `timeIncrement=1` to also get a daily breakdown: each node gains a `daily[]` array of per-day metrics (same fields as the aggregated `metrics`) in the same call. Use `dailyLevel` (`campaign` default, or `adset` / `ad`) to choose which levels carry the series. This replaces calling the tree once per day for per-campaign daily trends.  **Deleted objects stay in the tree.** Deleting an ad or a campaign is a soft delete: the Ad documents move to `status: cancelled` and are kept indefinitely, so their historical spend still counts toward the metrics of any date range they fall in. There is no pruning job and no retention window. Filter on `status` if your view should hide them, but do that after reading the totals, not before. 
+Shows your campaigns with their ad sets and ads nested inside, with spend and results rolled up at every level.  Uses a two-stage aggregation: ads are grouped into ad sets, then ad sets into campaigns. Metrics are computed over an optional date range, then rolled up from ad level to ad set and campaign levels. Pagination is at the campaign level. Ads without a campaign or ad set ID are grouped into synthetic \"Ungrouped\" buckets. If no date range is provided, defaults to the last 90 days. Date range is capped at 730 days max.  Pass `timeIncrement=1` to also get a daily breakdown: each node gains a `daily[]` array of per-day metrics (same fields as the aggregated `metrics`) in the same call. Use `dailyLevel` (`campaign` default, or `adset` / `ad`) to choose which levels carry the series. This replaces calling the tree once per day for per-campaign daily trends.  **Deleted objects stay in the tree.** Deleting an ad or a campaign is a soft delete: the Ad documents move to `status: cancelled` and are kept indefinitely, so their historical spend still counts toward the metrics of any date range they fall in. There is no pruning job and no retention window. Filter on `status` if your view should hide them, but do that after reading the totals, not before. 
 
 ### Examples
 
@@ -1901,7 +1901,7 @@ end
 
 Get daily account metrics
 
-Returns daily aggregate metrics across all ads in a SocialAccount as a single time series, one row per calendar day in the requested range. Use this for dashboards that draw a daily-spend or daily-conversions chart, instead of calling `/v1/ads/tree` once per day.  `accountId` is required. The lookup is sibling-expanded so passing the `metaads` ID also includes ads under the linked `facebook` / `instagram` posting account (and vice-versa), the same convention as `/v1/ads/tree` and `/v1/ads`.  Date range defaults to the last 90 days. Capped at 730 days. Ranges older than the ingested history return a `202` immediately with the covered part and `backfillPending: true` while the rest is backfilled in the background; repeat the request shortly until it returns 200 with full data.  With adAccountId set to a Google customer id this is the customer-level performance report (clicks, cost, impressions, conversions, all conversions per day). 
+Returns an ads account's spend, results and other metrics day by day, ready for a daily performance chart.  Returns daily aggregate metrics across all ads in a SocialAccount as a single time series, one row per calendar day in the requested range. Use this for dashboards that draw a daily-spend or daily-conversions chart, instead of calling `/v1/ads/tree` once per day.  `accountId` is required. The lookup is sibling-expanded so passing the `metaads` ID also includes ads under the linked `facebook` / `instagram` posting account (and vice-versa), the same convention as `/v1/ads/tree` and `/v1/ads`.  Date range defaults to the last 90 days. Capped at 730 days. Ranges older than the ingested history return a `202` immediately with the covered part and `backfillPending: true` while the rest is backfilled in the background; repeat the request shortly until it returns 200 with full data.  With adAccountId set to a Google customer id this is the customer-level performance report (clicks, cost, impressions, conversions, all conversions per day). 
 
 ### Examples
 
@@ -1980,7 +1980,7 @@ end
 
 Read a campaign's ad schedule (dayparting)
 
-The windows a Google campaign serves in, with the bid modifier on each, plus the criterion ids Google minted for them.  An EMPTY `schedule` is meaningful and is not a failed lookup: Google has no \"all day\" criterion, so a campaign with no ad schedule serves around the clock. `servesAroundTheClock` states that explicitly.  Set `includePerformance=true` to also get delivery split by day of week and by hour, which is the evidence for deciding what the schedule should be. It is one extra Google call segmented by both dimensions at once, so the two views always agree.  Google Ads only. The response carries `cachedAt` and `stale`, set when a quota-exhausted call falls back to the last-good copy instead of a live read. 
+Shows the days and hours a Google campaign runs and its bid adjustments, optionally with performance by day and hour.  The windows a Google campaign serves in, with the bid modifier on each, plus the criterion ids Google minted for them.  An EMPTY `schedule` is meaningful and is not a failed lookup: Google has no \"all day\" criterion, so a campaign with no ad schedule serves around the clock. `servesAroundTheClock` states that explicitly.  Set `includePerformance=true` to also get delivery split by day of week and by hour, which is the evidence for deciding what the schedule should be. It is one extra Google call segmented by both dimensions at once, so the two views always agree.  Google Ads only. The response carries `cachedAt` and `stale`, set when a quota-exhausted call falls back to the last-good copy instead of a live read. 
 
 ### Examples
 
@@ -2061,7 +2061,7 @@ end
 
 Read a campaign's current bidding
 
-Read of the campaign's bidding strategy on Google, cached for the quota window, for pre-filling the bid strategy block before a PUT to /v1/ads/campaigns/{campaignId}. Google Ads only; `platform` is required and rejected when it is anything else, since a `campaignId` is not globally unique. The response carries `cachedAt` and `stale`, set when a quota-exhausted call falls back to the last-good copy instead of a live read.  Maps Google's bidding strategy onto the same triplet PUT accepts: `LOWEST_COST_WITHOUT_CAP` (Maximize Conversions, no target), `COST_CAP` + `bidAmount` (Target CPA), `LOWEST_COST_WITH_MIN_ROAS` + `roasAverageFloor` (Target ROAS), `LOWEST_COST_WITH_BID_CAP` + `bidAmount` (Maximize Clicks with a CPC ceiling). A campaign on a portfolio strategy returns `portfolio` (id + name) and `bidSpec.portfolioBidStrategyId` instead of the triplet. Anything else (Manual CPC, Target Impression Share, ...) returns `bidSpec: null`; show `biddingStrategyType` as-is. 
+Shows a Google campaign's current bidding strategy and targets, ready to prefill an edit.  Read of the campaign's bidding strategy on Google, cached for the quota window, for pre-filling the bid strategy block before a PUT to /v1/ads/campaigns/{campaignId}. Google Ads only; `platform` is required and rejected when it is anything else, since a `campaignId` is not globally unique. The response carries `cachedAt` and `stale`, set when a quota-exhausted call falls back to the last-good copy instead of a live read.  Maps Google's bidding strategy onto the same triplet PUT accepts: `LOWEST_COST_WITHOUT_CAP` (Maximize Conversions, no target), `COST_CAP` + `bidAmount` (Target CPA), `LOWEST_COST_WITH_MIN_ROAS` + `roasAverageFloor` (Target ROAS), `LOWEST_COST_WITH_BID_CAP` + `bidAmount` (Maximize Clicks with a CPC ceiling). A campaign on a portfolio strategy returns `portfolio` (id + name) and `bidSpec.portfolioBidStrategyId` instead of the triplet. Anything else (Manual CPC, Target Impression Share, ...) returns `bidSpec: null`; show `biddingStrategyType` as-is. 
 
 ### Examples
 
@@ -2140,7 +2140,7 @@ end
 
 Get campaign conversion goals
 
-A Google campaign's conversion goals (CampaignConversionGoal, `biddable` per category and origin) and its goal config (ConversionGoalCampaignConfig): `goalConfigLevel` CUSTOMER means the campaign follows the account-default goals, CAMPAIGN means it uses its own goals or `customConversionGoalId`.
+Shows which conversion goals a Google campaign bids on, and whether it follows the account defaults or uses its own or a custom goal.  A Google campaign's conversion goals (CampaignConversionGoal, `biddable` per category and origin) and its goal config (ConversionGoalCampaignConfig): `goalConfigLevel` CUSTOMER means the campaign follows the account-default goals, CAMPAIGN means it uses its own goals or `customConversionGoalId`.
 
 ### Examples
 
@@ -2209,7 +2209,7 @@ end
 
 Read a Google campaign's device, location, excluded location, and language targeting
 
-Google Ads compliance requires geo, language, budget, and bidding targeting set at creation to stay editable afterwards; this reads the campaign state so an integrator can build an editor around it. Cached for the quota window (10 minutes fresh, up to 7 days last-good), not always a live read. Google only; every other platform returns 501.  `devices` lists the device criteria the campaign carries, which depends on its channel: Search campaigns have MOBILE, DESKTOP and TABLET, Display campaigns also have CONNECTED_TV. `bidModifier` is Google's bid adjustment for that device, `null` when it has none, and `0` when the device is switched off; `included` is false for exactly that case.  `excludedLocations` lists the campaign's negative location criteria (the places it never serves in). `locations` still lists every location criterion, each flagged with `negative`, so a client reading the targeted set filters `negative: false`. 
+Shows a Google campaign's device, location, excluded location and language targeting so you can build an editor for it.  Google Ads compliance requires geo, language, budget, and bidding targeting set at creation to stay editable afterwards; this reads the campaign state so an integrator can build an editor around it. Cached for the quota window (10 minutes fresh, up to 7 days last-good), not always a live read. Google only; every other platform returns 501.  `devices` lists the device criteria the campaign carries, which depends on its channel: Search campaigns have MOBILE, DESKTOP and TABLET, Display campaigns also have CONNECTED_TV. `bidModifier` is Google's bid adjustment for that device, `null` when it has none, and `0` when the device is switched off; `included` is false for exactly that case.  `excludedLocations` lists the campaign's negative location criteria (the places it never serves in). `locations` still lists every location criterion, each flagged with `negative`, so a client reading the targeted set filters `negative: false`. 
 
 ### Examples
 
@@ -2282,7 +2282,7 @@ end
 
 Get a Performance Max asset group
 
-One asset group with its linked assets, ad strength, primary status and listing-group tree. Uses a 10-minute cache, served stale when Google quota is exhausted; any write below clears it.
+Shows one Performance Max asset group with its linked assets, ad strength, status and product targeting tree.  One asset group with its linked assets, ad strength, primary status and listing-group tree. Uses a 10-minute cache, served stale when Google quota is exhausted; any write below clears it. 
 
 ### Examples
 
@@ -2353,7 +2353,7 @@ end
 
 List campaigns
 
-Returns campaigns as virtual aggregations over ad documents grouped by platform campaign ID. Metrics (spend, impressions, clicks, etc.) are summed across all ads in each campaign. Campaign status is derived from child ad statuses (active > pending_review > paused > error > completed > cancelled > rejected). Google campaign budgets include amountMicros, explicitlyShared, resourceName and deliveryMethod after the next successful sync. This endpoint does not fetch Google live.  **Status freshness.** `status`, `configuredStatus`, `platformStatus`, `platformAdSetStatus` and `platformCampaignStatus` are the values Zernio last stored. Background sync refreshes them, typically within 15 to 60 minutes (Google up to about 3 hours), and ended or long-paused objects may be refreshed less often. Zernio's own status writes re-read the switches they change. A change made in the platform's own ads manager therefore shows up here only after the next sync. Controllers that act on a switch should pass `live=true`, which reads the switches from the platform now, stores them, and returns `statusReadAt` (null when the read failed and the stored values were returned). With `live=true` (which needs `limit` of 20 or less) each returned campaign's own switch (`platformCampaignStatus`) is read live. Live reads cover TikTok, Meta, Google and OpenAI. The rolled-up `status` is not re-derived by a live read.  **Budgets here are SYNCED**, never read live, including with `live=true`. To read the current budgets (and status) of every campaign and ad set of a Meta ad account live in one call, for example as a pre-write spend gate, use GET /v1/ads/accounts/live. 
+Lists your campaigns with spend, impressions and clicks summed across their ads, plus each campaign's status and budget.  Returns campaigns as virtual aggregations over ad documents grouped by platform campaign ID. Metrics (spend, impressions, clicks, etc.) are summed across all ads in each campaign. Campaign status is derived from child ad statuses (active > pending_review > paused > error > completed > cancelled > rejected). Google campaign budgets include amountMicros, explicitlyShared, resourceName and deliveryMethod after the next successful sync. This endpoint does not fetch Google live.  **Status freshness.** `status`, `configuredStatus`, `platformStatus`, `platformAdSetStatus` and `platformCampaignStatus` are the values Zernio last stored. Background sync refreshes them, typically within 15 to 60 minutes (Google up to about 3 hours), and ended or long-paused objects may be refreshed less often. Zernio's own status writes re-read the switches they change. A change made in the platform's own ads manager therefore shows up here only after the next sync. Controllers that act on a switch should pass `live=true`, which reads the switches from the platform now, stores them, and returns `statusReadAt` (null when the read failed and the stored values were returned). With `live=true` (which needs `limit` of 20 or less) each returned campaign's own switch (`platformCampaignStatus`) is read live. Live reads cover TikTok, Meta, Google and OpenAI. The rolled-up `status` is not re-derived by a live read.  **Budgets here are SYNCED**, never read live, including with `live=true`. To read the current budgets (and status) of every campaign and ad set of a Meta ad account live in one call, for example as a pre-write spend gate, use GET /v1/ads/accounts/live. 
 
 ### Examples
 
@@ -2454,7 +2454,7 @@ end
 
 List ad-group assets
 
-Lists directly attached Google assets. Fresh reads are cached for 10 minutes; exhausted quota may return the last successful read with stale=true. Inherited assets are not included.
+Lists the sitelinks, callouts, snippets and images attached directly to a Google ad group, not those inherited from the campaign or account.  Fresh reads are cached for 10 minutes; exhausted quota may return the last successful read with stale=true. Inherited assets are not included. 
 
 ### Examples
 
@@ -2531,7 +2531,7 @@ end
 
 List Search keywords
 
-Returns the Google Search keyword criteria (positive and negative) synced from connected Google Ads accounts, one row per ad-group keyword. Refreshed about once a day per Google Ads customer (the keyword sweep rides the ads discovery pass on a slower slot), so keywords added on Google can take up to a day to appear. A customer synced for the first time is populated on the next discovery pass rather than waiting for its daily slot, and connecting an account or triggering a manual sync refreshes it immediately. Campaign-level negative keywords are not included; only ad-group-level criteria are. 
+Lists the keywords and negative keywords in your Google Search ad groups.  Returns the Google Search keyword criteria (positive and negative) synced from connected Google Ads accounts, one row per ad-group keyword. Refreshed about once a day per Google Ads customer (the keyword sweep rides the ads discovery pass on a slower slot), so keywords added on Google can take up to a day to appear. A customer synced for the first time is populated on the next discovery pass rather than waiting for its daily slot, and connecting an account or triggering a manual sync refreshes it immediately. Campaign-level negative keywords are not included; only ad-group-level criteria are. 
 
 ### Examples
 
@@ -2622,7 +2622,7 @@ end
 
 List ad sets
 
-Ad sets (Google ad groups) synced for the connection, optionally filtered by platform and campaignId. Reads the `ad_sets` table directly, independent of the `ads` rollup GET /v1/ads/tree uses, so a newly created standalone ad group with no ad yet (POST /v1/ads/ad-sets, Google only) is visible here even though it is invisible in the tree until an ad joins it via `adSetId` on POST /v1/ads/create. Returns at most 500 rows, newest first.  **This list is SYNCED, not live** (refreshed by background sync, see below). Each row also carries the ad set's config (`targeting`, `bidStrategy`, `bidAmount`, `optimizationGoal`, `billingEvent`, `promotedObject`), so `?accountId=...&adAccountId=act_...` returns the config of every ad set of an ad account in one call, without a `campaignId`. To read budgets, status, targeting, promoted object and bid strategy of every campaign and ad set of a Meta ad account LIVE in one call (for example as a pre-write spend gate), use GET /v1/ads/accounts/live instead.  **Status freshness.** `status`, `configuredStatus`, `platformStatus`, `platformAdSetStatus` and `platformCampaignStatus` are the values Zernio last stored. Background sync refreshes them, typically within 15 to 60 minutes (Google up to about 3 hours), and ended or long-paused objects may be refreshed less often. Zernio's own status writes re-read the switches they change. A change made in the platform's own ads manager therefore shows up here only after the next sync. Controllers that act on a switch should pass `live=true`, which reads the switches from the platform now, stores them, and returns `statusReadAt` (null when the read failed and the stored values were returned). With `live=true` (which needs a `campaignId` or `adSetId` filter) each listed ad set's own switch (`platformAdSetStatus`) is read live, for the first 20 rows; later rows keep their synced value with `statusReadAt: null`. Live reads cover TikTok, Meta, Google and OpenAI. The rolled-up `status` is not re-derived by a live read. On TikTok the same live read also returns the ad group's applied `optimizationGoal` and `billingEvent`, as TikTok's adgroup/get reports them, so you can verify the goal TikTok applied rather than the one you requested. It also returns `nativeSettings`: TikTok's own adgroup/get record for the ad group, verbatim and read in that same call (budget, budget_mode, schedule, placements, locations, ages, gender, languages, interests, actions, audiences and exclusions), plus the advertiser's currency and timezone. TikTok's `schedule_start_time` / `schedule_end_time` are UTC wall clocks (\"YYYY-MM-DD HH:MM:SS\", no offset); Ads Manager displays them in `advertiser_timezone`. `configReadAt` says when it was read; it is null on every row whose native settings were not read now, so never treat a null as a match. These are read, not retained create payloads, and are not stored.
+Lists ad sets (ad groups on Google) with their targeting, bidding and optimization settings, optionally for one campaign.  Ad sets (Google ad groups) synced for the connection, optionally filtered by platform and campaignId. Reads the `ad_sets` table directly, independent of the `ads` rollup GET /v1/ads/tree uses, so a newly created standalone ad group with no ad yet (POST /v1/ads/ad-sets, Google only) is visible here even though it is invisible in the tree until an ad joins it via `adSetId` on POST /v1/ads/create. Returns at most 500 rows, newest first.  **This list is SYNCED, not live** (refreshed by background sync, see below). Each row also carries the ad set's config (`targeting`, `bidStrategy`, `bidAmount`, `optimizationGoal`, `billingEvent`, `promotedObject`), so `?accountId=...&adAccountId=act_...` returns the config of every ad set of an ad account in one call, without a `campaignId`. To read budgets, status, targeting, promoted object and bid strategy of every campaign and ad set of a Meta ad account LIVE in one call (for example as a pre-write spend gate), use GET /v1/ads/accounts/live instead.  **Status freshness.** `status`, `configuredStatus`, `platformStatus`, `platformAdSetStatus` and `platformCampaignStatus` are the values Zernio last stored. Background sync refreshes them, typically within 15 to 60 minutes (Google up to about 3 hours), and ended or long-paused objects may be refreshed less often. Zernio's own status writes re-read the switches they change. A change made in the platform's own ads manager therefore shows up here only after the next sync. Controllers that act on a switch should pass `live=true`, which reads the switches from the platform now, stores them, and returns `statusReadAt` (null when the read failed and the stored values were returned). With `live=true` (which needs a `campaignId` or `adSetId` filter) each listed ad set's own switch (`platformAdSetStatus`) is read live, for the first 20 rows; later rows keep their synced value with `statusReadAt: null`. Live reads cover TikTok, Meta, Google and OpenAI. The rolled-up `status` is not re-derived by a live read. On TikTok the same live read also returns the ad group's applied `optimizationGoal` and `billingEvent`, as TikTok's adgroup/get reports them, so you can verify the goal TikTok applied rather than the one you requested. It also returns `nativeSettings`: TikTok's own adgroup/get record for the ad group, verbatim and read in that same call (budget, budget_mode, schedule, placements, locations, ages, gender, languages, interests, actions, audiences and exclusions), plus the advertiser's currency and timezone. TikTok's `schedule_start_time` / `schedule_end_time` are UTC wall clocks (\"YYYY-MM-DD HH:MM:SS\", no offset); Ads Manager displays them in `advertiser_timezone`. `configReadAt` says when it was read; it is null on every row whose native settings were not read now, so never treat a null as a match. These are read, not retained create payloads, and are not stored.
 
 ### Examples
 
@@ -2703,7 +2703,7 @@ end
 
 List ads
 
-Returns a paginated list of ads with metrics computed over an optional date range. Use source=all to include externally-synced ads from platform ad managers. If no date range is provided, defaults to the last 90 days. Date range is capped at 730 days max.  To find the Zernio ad behind a comment you see in Meta Business Manager, filter by platformAdId (the Meta ad ID), effectiveObjectStoryId (Facebook), or effectiveInstagramMediaId (Instagram). Those are the post/media the ad's engagement lives on, and are also returned on each ad's `creative` object. Then call GET /v1/ads/{adId}/comments with the returned ad id. 
+Lists your ads with their performance over a date range, including ads created directly in each platform's ads manager.  Returns a paginated list of ads with metrics computed over an optional date range. Use source=all to include externally-synced ads from platform ad managers. If no date range is provided, defaults to the last 90 days. Date range is capped at 730 days max.  To find the Zernio ad behind a comment you see in Meta Business Manager, filter by platformAdId (the Meta ad ID), effectiveObjectStoryId (Facebook), or effectiveInstagramMediaId (Instagram). Those are the post/media the ad's engagement lives on, and are also returned on each ad's `creative` object. Then call GET /v1/ads/{adId}/comments with the returned ad id. 
 
 ### Examples
 
@@ -2804,7 +2804,7 @@ end
 
 List portfolio bid strategies
 
-Bidding strategy report: type, status, campaign count, clicks, cost, cost per conversion, impressions, average CPC and conversions over the date range (default last 30 days). Reads Google's `bidding_strategy` resource, cached for the quota window. Draws on the shared Google Ads operations budget. The response carries `cachedAt` and `stale`, set when a quota-exhausted call falls back to the last-good copy instead of a live read.
+Lists a Google Ads account's shared bid strategies with how many campaigns use each and how they performed.  Bidding strategy report: type, status, campaign count, clicks, cost, cost per conversion, impressions, average CPC and conversions over the date range (default last 30 days). Reads Google's `bidding_strategy` resource, cached for the quota window. Draws on the shared Google Ads operations budget. The response carries `cachedAt` and `stale`, set when a quota-exhausted call falls back to the last-good copy instead of a live read.
 
 ### Examples
 
@@ -2883,7 +2883,7 @@ end
 
 List campaign assets
 
-Lists directly attached Google assets. Fresh reads are cached for 10 minutes; exhausted quota may return the last successful read with stale=true. Inherited assets are not included.
+Lists the sitelinks, callouts and other assets attached directly to a Google campaign.  Fresh reads are cached for 10 minutes; exhausted quota may return the last successful read with stale=true. Inherited assets are not included. 
 
 ### Examples
 
@@ -2960,7 +2960,7 @@ end
 
 List campaign negative lists
 
-Returns shared negative keyword lists attached to the campaign, separate from campaign-level negative keywords. Google Ads shared negative keyword lists (shared_set type NEGATIVE_KEYWORDS). Reads are cached for 10 minutes; quota exhaustion may return the last successful result for up to 7 days with stale=true. Customer selection is limited to this connection and its account scope.
+Lists the shared negative keyword lists attached to a Google campaign, separate from its own campaign-level negatives.  Google Ads shared negative keyword lists (shared_set type NEGATIVE_KEYWORDS). Reads are cached for 10 minutes; quota exhaustion may return the last successful result for up to 7 days with stale=true. Customer selection is limited to this connection and its account scope. 
 
 ### Examples
 
@@ -3033,7 +3033,7 @@ end
 
 List campaign-level negative keywords
 
-Returns the campaign-level negative keywords (`campaign_criterion.negative`), distinct from the ad-group-level negatives under `GET /v1/ads/keywords`. Cached for the quota window (not synced to Postgres), and gated by the shared Google Ads operations budget like every other on-demand Google surface. The response carries `cachedAt` and `stale`, set when a quota-exhausted call falls back to the last-good copy instead of a live read.  The platform is always discovered from the campaign itself; a non-Google campaign returns 501 rather than 404, whether or not `platform` was passed. 
+Lists the negative keywords set on a whole Google campaign, separate from ad group negatives.  Returns the campaign-level negative keywords (`campaign_criterion.negative`), distinct from the ad-group-level negatives under `GET /v1/ads/keywords`. Cached for the quota window (not synced to Postgres), and gated by the shared Google Ads operations budget like every other on-demand Google surface. The response carries `cachedAt` and `stale`, set when a quota-exhausted call falls back to the last-good copy instead of a live read.  The platform is always discovered from the campaign itself; a non-Google campaign returns 501 rather than 404, whether or not `platform` was passed. 
 
 ### Examples
 
@@ -3106,7 +3106,7 @@ end
 
 List Performance Max asset groups
 
-Read Performance Max asset groups and their linked text, image and YouTube assets. campaignId is the platform campaign id returned by creation or the campaign list. The campaign must be visible to the caller. Uses a 10-minute cache, with the last successful response served as stale when Google quota is exhausted. Removed groups and asset links are excluded. Campaign-level brand assets on campaigns with brand guidelines enabled are not included.
+Lists a Performance Max campaign's asset groups with their linked text, image and YouTube assets.  campaignId is the platform campaign id returned by creation or the campaign list. The campaign must be visible to the caller. Uses a 10-minute cache, with the last successful response served as stale when Google quota is exhausted. Removed groups and asset links are excluded. Campaign-level brand assets on campaigns with brand guidelines enabled are not included. 
 
 ### Examples
 
@@ -3175,7 +3175,7 @@ end
 
 List Google Ads recommendations
 
-Google's optimization recommendations for one ad account: type, estimated impact (base vs potential metrics, cost in account currency units), the campaign, ad group or budget they target, and the type-specific payload Google returns (`details`, in Google's own shape with micros). Filter by campaignId and types. Cached for 10 minutes and cleared by apply or dismiss; served stale when Google quota is exhausted.
+Lists Google's optimization suggestions for an ad account, with the estimated impact of each.  Google's optimization recommendations for one ad account: type, estimated impact (base vs potential metrics, cost in account currency units), the campaign, ad group or budget they target, and the type-specific payload Google returns (`details`, in Google's own shape with micros). Filter by campaignId and types. Cached for 10 minutes and cleared by apply or dismiss; served stale when Google quota is exhausted. 
 
 ### Examples
 
@@ -3254,7 +3254,7 @@ end
 
 List shared budgets
 
-Lists the Google Ads customer's shared campaign budgets (`campaign_budget.explicitly_shared` = true, not removed), with how many campaigns use each. Move a campaign onto one with `sharedBudgetId` on PUT /v1/ads/campaigns/{campaignId}. Google only; other platforms return 501. 
+Lists a Google Ads account's shared budgets and how many campaigns draw from each.  Lists the Google Ads customer's shared campaign budgets (`campaign_budget.explicitly_shared` = true, not removed), with how many campaigns use each. Move a campaign onto one with `sharedBudgetId` on PUT /v1/ads/campaigns/{campaignId}. Google only; other platforms return 501. 
 
 ### Examples
 
@@ -3327,7 +3327,7 @@ end
 
 Remove ad-group assets
 
-Removes the specified attachments only. Google assets cannot be deleted. Other attachments remain. assetResourceNames is retained for compatibility. Fields go in the query string. A JSON body with the same fields is also accepted and, when sent, the query string is ignored.
+Detaches assets from a Google ad group without deleting them, so they stay in the account and on anything else they are attached to.  Google assets cannot be deleted. Other attachments remain. assetResourceNames is retained for compatibility. Fields go in the query string. A JSON body with the same fields is also accepted and, when sent, the query string is ignored. 
 
 ### Examples
 
@@ -3408,7 +3408,7 @@ end
 
 Remove a Search keyword
 
-Removes one keyword criterion (positive or negative) from its ad group (M.140).
+Removes a single keyword or negative keyword from a Google Search ad group.  Removes one keyword criterion (positive or negative) from its ad group (M.140). 
 
 ### Examples
 
@@ -3477,7 +3477,7 @@ end
 
 Remove campaign assets
 
-Removes the specified attachments only. Google assets cannot be deleted. Other attachments remain. assetResourceNames is retained for compatibility. Fields go in the query string. A JSON body with the same fields is also accepted and, when sent, the query string is ignored.
+Detaches assets from a Google campaign without deleting them, so they stay in the account and on anything else they are attached to.  Google assets cannot be deleted. Other attachments remain. assetResourceNames is retained for compatibility. Fields go in the query string. A JSON body with the same fields is also accepted and, when sent, the query string is ignored. 
 
 ### Examples
 
@@ -3558,7 +3558,7 @@ end
 
 Remove a Performance Max asset group
 
-Removes the asset group on Google (status REMOVED, not reversible). Pass validateOnly=true to validate without removing.
+Permanently removes an asset group from a Performance Max campaign (this cannot be undone).  Removes the asset group on Google (status REMOVED, not reversible). Pass validateOnly=true to validate without removing. 
 
 ### Examples
 
@@ -3633,7 +3633,7 @@ end
 
 Replace campaign negative lists
 
-Sets the full desired set of shared negative keyword list associations on this campaign. Send listIds=[] to detach all negative keyword lists. Only campaign_shared_set links are changed; the lists and their keywords are preserved. Every list must belong to the campaign customer and have type NEGATIVE_KEYWORDS.
+Sets exactly which shared negative keyword lists apply to a Google campaign, or detaches them all, without changing the lists themselves.  Send listIds=[] to detach all negative keyword lists. Only campaign_shared_set links are changed; the lists and their keywords are preserved. Every list must belong to the campaign customer and have type NEGATIVE_KEYWORDS. 
 
 ### Examples
 
@@ -3704,7 +3704,7 @@ end
 
 Replace campaign-level negative keywords
 
-Replaces the FULL set of campaign-level negative keywords (C.270): the desired list is diffed against what Google already has, and the difference is applied as one `create`/`remove` mutate. Send an empty array to clear every campaign negative.  The platform is always discovered from the campaign itself; a non-Google campaign returns 501 rather than 404, whether or not `platform` was sent. 
+Replaces a Google campaign's full list of campaign-level negative keywords, or clears it.  Replaces the FULL set of campaign-level negative keywords (C.270): the desired list is diffed against what Google already has, and the difference is applied as one `create`/`remove` mutate. Send an empty array to clear every campaign negative.  The platform is always discovered from the campaign itself; a non-Google campaign returns 501 rather than 404, whether or not `platform` was sent. 
 
 ### Examples
 
@@ -3775,7 +3775,7 @@ end
 
 Replace an asset group's listing-group tree
 
-Replace the product (listing-group) tree of a Performance Max retail asset group. The current tree is removed and the new one created in one atomic request. Read the current tree with GET on the asset group. Requires a campaign linked to Merchant Center; other campaigns return 400 LISTING_SOURCE_NOT_ALLOWED from Google. validateOnly: true validates without writing.
+Replaces the product targeting tree of a Performance Max retail asset group; only works on campaigns linked to Merchant Center.  Replace the product (listing-group) tree of a Performance Max retail asset group. The current tree is removed and the new one created in one atomic request. Read the current tree with GET on the asset group. Requires a campaign linked to Merchant Center; other campaigns return 400 LISTING_SOURCE_NOT_ALLOWED from Google. validateOnly: true validates without writing. 
 
 ### Examples
 
@@ -3848,7 +3848,7 @@ end
 
 Update ad
 
-Patch one or more fields on an ad. Status, budget, targeting, and creative changes are propagated to the platform.  Per-platform support: - **Meta** (Facebook + Instagram): all fields supported. - **TikTok**: status, budget, `name` (renames the ad), targeting (via `/v2/adgroup/update/`), and creative   (via `/v2/ad/update/` patch-style: `headline` is ignored, `body` becomes `ad_text`). - **Google**: status, budget, KEYWORD edits via `targeting.keywords` /   `targeting.negativeKeywords`, DEVICE bid adjustments via `targeting.devices`,   LOCATION edits via `targeting.locations` (or the equivalent top-level   `targeting.countries` / `regions` / `cities` / `zips` / `metros`), and LANGUAGE   edits via `targeting.languages`.   Each list you send becomes the FULL new set of its kind (criteria not in the   list are removed, except devices, which Google cannot remove and which are   switched off with a bid modifier of 0 instead); a kind left out is untouched.   Any other `targeting` field   returns 400: Google cannot mutate it post-create without recreating   the campaign. Creative edits are dispatched on the ad's `advertisingChannelType`,   and every supported field replaces a whole set; a field you omit is preserved.   - **Search**: top-level `headlines`, `descriptions` and `finalUrls`. Use 3-15 headlines     (1-30 characters) and 2-4 descriptions (1-90 characters). Omit an asset to remove it;     omit pinnedField on an included asset to unpin it. Updates do not pad or truncate text.     The legacy creative fields remain unsupported.   - **Display**: top-level `headlines` (1-5, no pinnedField, display ads have no pinned     positions), `descriptions` (1-5) and `finalUrls`, plus `creative.longHeadline`,     `creative.businessName`, `creative.imageUrl` (the landscape marketing image) and     `creative.squareImageUrl`. Each image URL is uploaded as a new Google asset and the ad     is pointed at it; Google assets are immutable, so the previous asset stays in the     account's asset library.   - **Performance Max**: top-level `assetGroup`, which swaps asset roles on the ad's asset     group. The other creative fields return 422 for this channel, and `assetGroup` returns     422 on any other channel.   - **Demand Gen**: top-level `demandGen` (see GoogleDemandGenUpdate): creative, channel     and audience changes in one atomic Google request. The other creative fields return     422 for this channel. `targeting` takes locations, languages, `locationTargetingType`     and `devices`; locations and languages are written to the ad's ad group, which     Demand Gen requires (campaigns migrated from Discovery that still target on the     campaign keep being written there), so they apply to every ad in that ad group. - **LinkedIn**: status, budget, targeting (countries or regions, excludedLocations (countries),   the B2B facets, and audience segments; applied to the LinkedIn Campaign via   PARTIAL_UPDATE, and REPLACES the campaign's entire targetingCriteria, not a merge),   and creative (uploads new media, creates a replacement inline creative on the same   campaign, pauses the old one). - **Pinterest / X / OpenAI Ads**: status + budget only. Sending   `targeting` or `creative` returns 501 with code `unsupported_platform_operation`.   OpenAI Ads budget is the campaign's spend cap, daily or lifetime (see `budget.type` below).  **Google location and language replacement:** locations, languages and devices are campaign-level criteria on Google, so these edits apply to every ad group and ad in the ad's campaign. Send the complete list you want to keep. Zernio diffs it against the campaign's live criteria and sends the removes and the creates in ONE `googleAds:mutate`, so the campaign is never left with a half-applied set; criteria already in the list keep their criterion ID and history. Excluded (negative) locations are left untouched. An empty location list returns 400 (a Google campaign with no location criteria targets every country, which is never what \"remove my locations\" means, so omit the field instead). Send either `targeting.locations` or the top-level geo fields, not both: mixing them returns 400.  **Google radius targeting:** `customLocations` is editable and is replaced the same way, but as its OWN set. Google models a place (LOCATION) and a point plus radius (PROXIMITY) as different criterion types, so the two are independent: sending `customLocations` replaces every radius and leaves the cities and countries alone, and sending places replaces those and leaves the radius alone. Send `customLocations: []` to drop radius targeting entirely. A circle you re-send unchanged keeps its criterion ID rather than being removed and recreated.  **Google keyword replacement:** These edits affect the ad's entire ad group, including sibling ads. Positive (`targeting.keywords`) and negative (`targeting.negativeKeywords`) sets are independent: omit a field to leave that set unchanged, or send `[]` to remove every keyword of that kind.  Zernio compares each supplied set with Google's live criteria by case-insensitive keyword text and match type. A matching criterion is left untouched, retaining its criterion ID, enabled/paused status, keyword-level bid overrides, labels, and criterion-associated history/statistics. Zernio does not reset its quality score; Google continues to calculate scores and statistics normally. Text comparison does not trim whitespace.  A bare string or an object without `matchType` means `broad`, not the existing criterion's match type. For example, resending an existing `{ \"text\": \"plumber\", \"matchType\": \"exact\" }` preserves it; sending `\"plumber\"` instead removes that EXACT criterion and requests a BROAD one. Changing text or match type removes criteria no longer requested and creates any missing criteria. New criteria get new IDs and do not inherit removed criteria's bid overrides, labels, or history. Historical reporting for a removed criterion is not transferred to its replacement.  To add keywords without replacing a set, use [POST /v1/ads/keywords](https://docs.zernio.com/ad-campaigns/add-ad-keywords). Use `PATCH /v1/ads/keywords/{keywordId}` to pause/enable one keyword, or `DELETE /v1/ads/keywords/{keywordId}` to remove it. 
+Changes an ad's status, budget, targeting or creative and applies the change on the platform.  Patch one or more fields on an ad. Status, budget, targeting, and creative changes are propagated to the platform.  Per-platform support: - **Meta** (Facebook + Instagram): all fields supported. - **TikTok**: status, budget, `name` (renames the ad), targeting (via `/v2/adgroup/update/`), and creative   (via `/v2/ad/update/` patch-style: `headline` is ignored, `body` becomes `ad_text`). - **Google**: status, budget, KEYWORD edits via `targeting.keywords` /   `targeting.negativeKeywords`, DEVICE bid adjustments via `targeting.devices`,   LOCATION edits via `targeting.locations` (or the equivalent top-level   `targeting.countries` / `regions` / `cities` / `zips` / `metros`), and LANGUAGE   edits via `targeting.languages`.   Each list you send becomes the FULL new set of its kind (criteria not in the   list are removed, except devices, which Google cannot remove and which are   switched off with a bid modifier of 0 instead); a kind left out is untouched.   Any other `targeting` field   returns 400: Google cannot mutate it post-create without recreating   the campaign. Creative edits are dispatched on the ad's `advertisingChannelType`,   and every supported field replaces a whole set; a field you omit is preserved.   - **Search**: top-level `headlines`, `descriptions` and `finalUrls`. Use 3-15 headlines     (1-30 characters) and 2-4 descriptions (1-90 characters). Omit an asset to remove it;     omit pinnedField on an included asset to unpin it. Updates do not pad or truncate text.     The legacy creative fields remain unsupported.   - **Display**: top-level `headlines` (1-5, no pinnedField, display ads have no pinned     positions), `descriptions` (1-5) and `finalUrls`, plus `creative.longHeadline`,     `creative.businessName`, `creative.imageUrl` (the landscape marketing image) and     `creative.squareImageUrl`. Each image URL is uploaded as a new Google asset and the ad     is pointed at it; Google assets are immutable, so the previous asset stays in the     account's asset library.   - **Performance Max**: top-level `assetGroup`, which swaps asset roles on the ad's asset     group. The other creative fields return 422 for this channel, and `assetGroup` returns     422 on any other channel.   - **Demand Gen**: top-level `demandGen` (see GoogleDemandGenUpdate): creative, channel     and audience changes in one atomic Google request. The other creative fields return     422 for this channel. `targeting` takes locations, languages, `locationTargetingType`     and `devices`; locations and languages are written to the ad's ad group, which     Demand Gen requires (campaigns migrated from Discovery that still target on the     campaign keep being written there), so they apply to every ad in that ad group. - **LinkedIn**: status, budget, targeting (countries or regions, excludedLocations (countries),   the B2B facets, and audience segments; applied to the LinkedIn Campaign via   PARTIAL_UPDATE, and REPLACES the campaign's entire targetingCriteria, not a merge),   and creative (uploads new media, creates a replacement inline creative on the same   campaign, pauses the old one). - **Pinterest / X / OpenAI Ads**: status + budget only. Sending   `targeting` or `creative` returns 501 with code `unsupported_platform_operation`.   OpenAI Ads budget is the campaign's spend cap, daily or lifetime (see `budget.type` below).  **Google location and language replacement:** locations, languages and devices are campaign-level criteria on Google, so these edits apply to every ad group and ad in the ad's campaign. Send the complete list you want to keep. Zernio diffs it against the campaign's live criteria and sends the removes and the creates in ONE `googleAds:mutate`, so the campaign is never left with a half-applied set; criteria already in the list keep their criterion ID and history. Excluded (negative) locations are left untouched. An empty location list returns 400 (a Google campaign with no location criteria targets every country, which is never what \"remove my locations\" means, so omit the field instead). Send either `targeting.locations` or the top-level geo fields, not both: mixing them returns 400.  **Google radius targeting:** `customLocations` is editable and is replaced the same way, but as its OWN set. Google models a place (LOCATION) and a point plus radius (PROXIMITY) as different criterion types, so the two are independent: sending `customLocations` replaces every radius and leaves the cities and countries alone, and sending places replaces those and leaves the radius alone. Send `customLocations: []` to drop radius targeting entirely. A circle you re-send unchanged keeps its criterion ID rather than being removed and recreated.  **Google keyword replacement:** These edits affect the ad's entire ad group, including sibling ads. Positive (`targeting.keywords`) and negative (`targeting.negativeKeywords`) sets are independent: omit a field to leave that set unchanged, or send `[]` to remove every keyword of that kind.  Zernio compares each supplied set with Google's live criteria by case-insensitive keyword text and match type. A matching criterion is left untouched, retaining its criterion ID, enabled/paused status, keyword-level bid overrides, labels, and criterion-associated history/statistics. Zernio does not reset its quality score; Google continues to calculate scores and statistics normally. Text comparison does not trim whitespace.  A bare string or an object without `matchType` means `broad`, not the existing criterion's match type. For example, resending an existing `{ \"text\": \"plumber\", \"matchType\": \"exact\" }` preserves it; sending `\"plumber\"` instead removes that EXACT criterion and requests a BROAD one. Changing text or match type removes criteria no longer requested and creates any missing criteria. New criteria get new IDs and do not inherit removed criteria's bid overrides, labels, or history. Historical reporting for a removed criterion is not transferred to its replacement.  To add keywords without replacing a set, use [POST /v1/ads/keywords](https://docs.zernio.com/ad-campaigns/add-ad-keywords). Use `PATCH /v1/ads/keywords/{keywordId}` to pause/enable one keyword, or `DELETE /v1/ads/keywords/{keywordId}` to remove it. 
 
 ### Examples
 
@@ -3919,7 +3919,7 @@ end
 
 Update a campaign
 
-Campaign-level edits. Send at least one of `budget`, `bidStrategy`, `portfolioBidStrategyId`, `targetImpressionShare`, `manualCpc`, `networkSettings`, `trackingUrlTemplate`, `finalUrlSuffix`, `sharedBudgetId`, `name` or `platformSpecificData`. An unsupported field is always an error, never a silent drop.  | Body field | Meta | Google | Others | |---|---|---|---| | `bidStrategy` | Yes | Yes | 501 | | `bidAmount`, `roasAverageFloor` | 400 (ad-set level) | Yes | 400 | | `portfolioBidStrategyId` | 400 | Yes | 400 | | `targetImpressionShare` | 400 | Search only | 400 | | `manualCpc` | 400 | Search and Display | 400 | | `networkSettings` | 400 | Search only | 400 | | `trackingUrlTemplate`, `finalUrlSuffix` | 400 | Yes | 400 | | `sharedBudgetId` | 400 | Yes | 400 | | `budget` (CBO; ABO returns 409) | Yes | Daily only | OpenAI: daily or lifetime; others 501 | | `name` | Yes | 501 | 501 | | `platformSpecificData.spendCap` | Yes | 400 | 400 | | `accountId` (empty campaigns) | Yes | - | - |  Meta budget edits check the live campaign budget, so an older local ABO stamp cannot block a CBO campaign. A successful edit repairs local ad budget fields. A live ABO campaign still returns 409 with the ad-set budget endpoint.  On Google: `LOWEST_COST_WITHOUT_CAP` = Maximize Conversions, `COST_CAP` + `bidAmount` = Target CPA, `LOWEST_COST_WITH_MIN_ROAS` + `roasAverageFloor` = Target ROAS, `LOWEST_COST_WITH_BID_CAP` + `bidAmount` = Maximize Clicks with a CPC ceiling; `portfolioBidStrategyId` attaches a portfolio strategy instead (exclusive with `bidStrategy`). `targetImpressionShare` switches the campaign to Target impression share and `manualCpc: { maxCpc }` to Manual CPC (every ad group of the campaign gets `maxCpc` as its bid, in the same mutate); each is exclusive with every other strategy field. `networkSettings`, `trackingUrlTemplate` and `finalUrlSuffix` go out in one campaign update, each field sent written on its own leaf, so an omitted one keeps its current value; an empty string clears a URL field. Setting the standard triplet on a campaign that is currently on a PORTFOLIO strategy is rejected: detach it in Google Ads first, since it is shared across campaigns.  Google budget updates read the current budget before mutation. Shared budgets return 409 unless allowSharedBudgetUpdate=true is explicitly supplied, because the change affects every campaign using that budget. A budget counts as shared when Google flags it as shared or when more than one campaign uses it (older budgets can serve several campaigns without the flag). Unknown sharing state also returns 409.  `sharedBudgetId` (Google) moves the campaign onto a shared budget from GET /v1/ads/shared-budgets, which also needs `allowSharedBudgetUpdate: true` (409 otherwise) because the campaign then splits that budget with every campaign on it; `budget` cannot ride along. `sharedBudgetId: null` moves it back onto a new budget of its own, sized by `budget` (required, daily); the new budget and the switch go out in one atomic Google mutate. A campaign that already has its own budget returns 409 for null. The budget a campaign leaves is not removed. The response carries the budget the campaign now uses.  OpenAI Ads campaigns carry exactly one spend cap: `budget.type` daily or lifetime replaces whichever cap the campaign had, with a minimum of 1 in the ad account's currency (422 below it). Lifetime can switch to daily, but OpenAI never switches a daily cap back to lifetime (422).  `accountId` forwards the update straight to Meta for a campaign with zero ads, which would otherwise 404; the response then carries `updated: 0`. 
+Changes a campaign's settings, such as its name, budget, bidding or tracking URLs.  Send at least one of `budget`, `bidStrategy`, `portfolioBidStrategyId`, `targetImpressionShare`, `manualCpc`, `networkSettings`, `trackingUrlTemplate`, `finalUrlSuffix`, `sharedBudgetId`, `name` or `platformSpecificData`. An unsupported field is always an error, never a silent drop.  | Body field | Meta | Google | Others | |---|---|---|---| | `bidStrategy` | Yes | Yes | 501 | | `bidAmount`, `roasAverageFloor` | 400 (ad-set level) | Yes | 400 | | `portfolioBidStrategyId` | 400 | Yes | 400 | | `targetImpressionShare` | 400 | Search only | 400 | | `manualCpc` | 400 | Search and Display | 400 | | `networkSettings` | 400 | Search only | 400 | | `trackingUrlTemplate`, `finalUrlSuffix` | 400 | Yes | 400 | | `sharedBudgetId` | 400 | Yes | 400 | | `budget` (CBO; ABO returns 409) | Yes | Daily only | OpenAI: daily or lifetime; others 501 | | `name` | Yes | 501 | 501 | | `platformSpecificData.spendCap` | Yes | 400 | 400 | | `accountId` (empty campaigns) | Yes | - | - |  Meta budget edits check the live campaign budget, so an older local ABO stamp cannot block a CBO campaign. A successful edit repairs local ad budget fields. A live ABO campaign still returns 409 with the ad-set budget endpoint.  On Google: `LOWEST_COST_WITHOUT_CAP` = Maximize Conversions, `COST_CAP` + `bidAmount` = Target CPA, `LOWEST_COST_WITH_MIN_ROAS` + `roasAverageFloor` = Target ROAS, `LOWEST_COST_WITH_BID_CAP` + `bidAmount` = Maximize Clicks with a CPC ceiling; `portfolioBidStrategyId` attaches a portfolio strategy instead (exclusive with `bidStrategy`). `targetImpressionShare` switches the campaign to Target impression share and `manualCpc: { maxCpc }` to Manual CPC (every ad group of the campaign gets `maxCpc` as its bid, in the same mutate); each is exclusive with every other strategy field. `networkSettings`, `trackingUrlTemplate` and `finalUrlSuffix` go out in one campaign update, each field sent written on its own leaf, so an omitted one keeps its current value; an empty string clears a URL field. Setting the standard triplet on a campaign that is currently on a PORTFOLIO strategy is rejected: detach it in Google Ads first, since it is shared across campaigns.  Google budget updates read the current budget before mutation. Shared budgets return 409 unless allowSharedBudgetUpdate=true is explicitly supplied, because the change affects every campaign using that budget. A budget counts as shared when Google flags it as shared or when more than one campaign uses it (older budgets can serve several campaigns without the flag). Unknown sharing state also returns 409.  `sharedBudgetId` (Google) moves the campaign onto a shared budget from GET /v1/ads/shared-budgets, which also needs `allowSharedBudgetUpdate: true` (409 otherwise) because the campaign then splits that budget with every campaign on it; `budget` cannot ride along. `sharedBudgetId: null` moves it back onto a new budget of its own, sized by `budget` (required, daily); the new budget and the switch go out in one atomic Google mutate. A campaign that already has its own budget returns 409 for null. The budget a campaign leaves is not removed. The response carries the budget the campaign now uses.  OpenAI Ads campaigns carry exactly one spend cap: `budget.type` daily or lifetime replaces whichever cap the campaign had, with a minimum of 1 in the ad account's currency (422 below it). Lifetime can switch to daily, but OpenAI never switches a daily cap back to lifetime (422).  `accountId` forwards the update straight to Meta for a campaign with zero ads, which would otherwise 404; the response then carries `updated: 0`. 
 
 ### Examples
 
@@ -3990,7 +3990,7 @@ end
 
 Pause or resume a campaign
 
-Writes the campaign's own on/off switch and nothing else, on every platform (Meta, TikTok, Google, LinkedIn campaign group, Pinterest, X, ChatGPT (OpenAI)). Its ad sets and ads keep their own switches: pausing stops their delivery through the campaign, and resuming lets each of them deliver again only if its own switch is on. An ad set or ad you paused individually stays paused; resume it with PUT /v1/ads/ad-sets/{adSetId}/status or PUT /v1/ads/{adId}/status. See the Status model in the Ad Campaigns tag.  **Live read, then write.** The campaign's switch is read from the platform first. When that live read shows it already in the requested state nothing is written (`updated: 0`, `skipped: 1`, with the reason). Otherwise the switch is written (`updated: 1`), read back and stored, and the delivery status of the ads under it (up to 20) is re-read and stored, so an immediate GET returns what the platform now reports. A stored switch never skips a write, and when the platform cannot be read the write always goes out. On Meta the check reads the campaign's own `status`, so a delivery status such as `IN_PROCESS` or `WITH_ISSUES` does not force a write. 
+Pauses or resumes a campaign, leaving the on/off switches of its ad sets and ads as they are.  Writes the campaign's own on/off switch and nothing else, on every platform (Meta, TikTok, Google, LinkedIn campaign group, Pinterest, X, ChatGPT (OpenAI)). Its ad sets and ads keep their own switches: pausing stops their delivery through the campaign, and resuming lets each of them deliver again only if its own switch is on. An ad set or ad you paused individually stays paused; resume it with PUT /v1/ads/ad-sets/{adSetId}/status or PUT /v1/ads/{adId}/status. See the Status model in the Ad Campaigns tag.  **Live read, then write.** The campaign's switch is read from the platform first. When that live read shows it already in the requested state nothing is written (`updated: 0`, `skipped: 1`, with the reason). Otherwise the switch is written (`updated: 1`), read back and stored, and the delivery status of the ads under it (up to 20) is re-read and stored, so an immediate GET returns what the platform now reports. A stored switch never skips a write, and when the platform cannot be read the write always goes out. On Meta the check reads the campaign's own `status`, so a delivery status such as `IN_PROCESS` or `WITH_ISSUES` does not force a write. 
 
 ### Examples
 
@@ -4061,7 +4061,7 @@ end
 
 Update ad-group assets
 
-Edits existing Google assets in place. Send updates with assetResourceName and the fields to change. An asset is shared: changes affect every attachment using it. Omitted fields stay unchanged. The operation consumes the Google operations budget and invalidates affected cached lists.
+Edits a Google ad group's existing assets in place; the change applies everywhere that asset is used.  Send updates with assetResourceName and the fields to change. An asset is shared: changes affect every attachment using it. Omitted fields stay unchanged. The operation consumes the Google operations budget and invalidates affected cached lists. 
 
 ### Examples
 
@@ -4132,7 +4132,7 @@ end
 
 Pause or enable a Search keyword
 
-Changes `ad_group_criterion.status` for one keyword criterion (M.140). Negative keywords have no status on Google and cannot be paused or enabled. 
+Pauses or re-enables a single Google Search keyword.  Changes `ad_group_criterion.status` for one keyword criterion (M.140). Negative keywords have no status on Google and cannot be paused or enabled. 
 
 ### Examples
 
@@ -4203,7 +4203,7 @@ end
 
 Update an ad set
 
-Ad-set-level writes. Use this for ABO budget updates, ad-set-scoped pause/resume, bid-strategy edits, Meta value-rule-set attach/detach, and Meta-only post-launch delivery settings via `platformSpecificData`. At least one updatable field is required.  Value rule sets (Meta only, see `/v1/ads/value-rule-sets`): - ATTACH or REPLACE: send `valueRuleSetId`. Attachment is driven by the id's   presence, so `valueRulesApplied: true` is optional. Sending a different id   replaces the previous association; there is no separate replace call. - DETACH: send `valueRulesApplied: false` and OMIT `valueRuleSetId`. - Sending `valueRulesApplied: false` TOGETHER with `valueRuleSetId` returns 400   `mutually_exclusive_fields`. This is deliberate: Meta attaches the rule set   whenever `value_rule_set_id` is present, even with `value_rules_applied` false,   so echoing stored state while asking to detach would silently keep the bid   adjustments live. - Eligibility: only ad sets on `LOWEST_COST_WITHOUT_CAP` or `COST_CAP`. Meta   rejects the rest server-side. - Read back with `GET /v1/ads/ad-sets/{adSetId}?fields=value_rule_set_id`. Meta   does not document `value_rules_applied` as a readable ad-set field, so the   boolean cannot be read back.  Bid strategy compatibility (per Meta's spec): - `LOWEST_COST_WITHOUT_CAP`: no `bidAmount`, no `roasAverageFloor`. - `LOWEST_COST_WITH_BID_CAP` / `COST_CAP`: `bidAmount` REQUIRED (whole currency units). - `LOWEST_COST_WITH_MIN_ROAS`: `roasAverageFloor` REQUIRED (decimal multiplier, e.g. 2.0 = 2.0x ROAS). - Meta only: send `bidAmount` WITHOUT `bidStrategy` to change the cap amount on an ad set   under a COST_CAP / LOWEST_COST_WITH_BID_CAP parent campaign, leaving the strategy itself   (inherited from the campaign) untouched. `roasAverageFloor` without `bidStrategy` is   rejected (it has no meaning outside LOWEST_COST_WITH_MIN_ROAS).  Delivery settings are validated by Meta against the campaign objective; incompatible combinations (e.g. a billingEvent the optimization goal doesn't allow) surface as 400s from Meta.  When updating `budget` on an ABO campaign: if the parent campaign is CBO, the response is 409 with code BUDGET_LEVEL_MISMATCH. Route to PUT /v1/ads/campaigns/{campaignId} instead.  `status` behaves exactly as PUT /v1/ads/ad-sets/{adSetId}/status describes: only the ad set's own switch is written, its ads keep theirs, and `statusUpdated` / `statusSkipped` report whether that switch was written or a live read showed it already in place. 
+Changes an ad set's settings, such as its budget, bid strategy, status or delivery options.  Use this for ABO budget updates, ad-set-scoped pause/resume, bid-strategy edits, Meta value-rule-set attach/detach, and Meta-only post-launch delivery settings via `platformSpecificData`. At least one updatable field is required.  Value rule sets (Meta only, see `/v1/ads/value-rule-sets`): - ATTACH or REPLACE: send `valueRuleSetId`. Attachment is driven by the id's   presence, so `valueRulesApplied: true` is optional. Sending a different id   replaces the previous association; there is no separate replace call. - DETACH: send `valueRulesApplied: false` and OMIT `valueRuleSetId`. - Sending `valueRulesApplied: false` TOGETHER with `valueRuleSetId` returns 400   `mutually_exclusive_fields`. This is deliberate: Meta attaches the rule set   whenever `value_rule_set_id` is present, even with `value_rules_applied` false,   so echoing stored state while asking to detach would silently keep the bid   adjustments live. - Eligibility: only ad sets on `LOWEST_COST_WITHOUT_CAP` or `COST_CAP`. Meta   rejects the rest server-side. - Read back with `GET /v1/ads/ad-sets/{adSetId}?fields=value_rule_set_id`. Meta   does not document `value_rules_applied` as a readable ad-set field, so the   boolean cannot be read back.  Bid strategy compatibility (per Meta's spec): - `LOWEST_COST_WITHOUT_CAP`: no `bidAmount`, no `roasAverageFloor`. - `LOWEST_COST_WITH_BID_CAP` / `COST_CAP`: `bidAmount` REQUIRED (whole currency units). - `LOWEST_COST_WITH_MIN_ROAS`: `roasAverageFloor` REQUIRED (decimal multiplier, e.g. 2.0 = 2.0x ROAS). - Meta only: send `bidAmount` WITHOUT `bidStrategy` to change the cap amount on an ad set   under a COST_CAP / LOWEST_COST_WITH_BID_CAP parent campaign, leaving the strategy itself   (inherited from the campaign) untouched. `roasAverageFloor` without `bidStrategy` is   rejected (it has no meaning outside LOWEST_COST_WITH_MIN_ROAS).  Delivery settings are validated by Meta against the campaign objective; incompatible combinations (e.g. a billingEvent the optimization goal doesn't allow) surface as 400s from Meta.  When updating `budget` on an ABO campaign: if the parent campaign is CBO, the response is 409 with code BUDGET_LEVEL_MISMATCH. Route to PUT /v1/ads/campaigns/{campaignId} instead.  `status` behaves exactly as PUT /v1/ads/ad-sets/{adSetId}/status describes: only the ad set's own switch is written, its ads keep theirs, and `statusUpdated` / `statusSkipped` report whether that switch was written or a live read showed it already in place. 
 
 ### Examples
 
@@ -4274,7 +4274,7 @@ end
 
 Pause or resume a single ad set
 
-Ad-set-scoped pause/resume (doesn't touch sibling ad sets). Thin wrapper over PUT /v1/ads/ad-sets/{adSetId} for callers that only want the status toggle and prefer a symmetric URL to /v1/ads/campaigns/{campaignId}/status.  Writes the ad set's own on/off switch and nothing else, on every platform (Meta `configured_status`, TikTok ad group `operation_status`, Google ad group status, LinkedIn campaign, Pinterest ad group, X line item, ChatGPT (OpenAI) ad group). Its ads keep their own switches: an ad you paused individually stays paused when the ad set resumes. The campaign above is not touched either, so an ad set resumed under a paused campaign reads `status: paused` until the campaign is resumed too. See the Status model in the Ad Campaigns tag.  **Live read, then write.** The ad set's switch is read from the platform first. When that live read shows it already in the requested state nothing is written (`updated: 0`, `skipped: 1`, with the reason). Otherwise the switch is written (`updated: 1`), read back and stored, and the delivery status of its ads (up to 20) is re-read and stored, so an immediate GET returns what the platform now reports. A stored switch never skips a write, and when the platform cannot be read the write always goes out. On Meta the check reads the ad set's own `status`, so a repeated request skips even while `platformAdSetStatus` reads `CAMPAIGN_PAUSED`, `WITH_ISSUES` or `IN_PROCESS`. 
+Pauses or resumes one ad set without touching its campaign, its sibling ad sets or the switches of its ads.  Thin wrapper over PUT /v1/ads/ad-sets/{adSetId} for callers that only want the status toggle and prefer a symmetric URL to /v1/ads/campaigns/{campaignId}/status.  Writes the ad set's own on/off switch and nothing else, on every platform (Meta `configured_status`, TikTok ad group `operation_status`, Google ad group status, LinkedIn campaign, Pinterest ad group, X line item, ChatGPT (OpenAI) ad group). Its ads keep their own switches: an ad you paused individually stays paused when the ad set resumes. The campaign above is not touched either, so an ad set resumed under a paused campaign reads `status: paused` until the campaign is resumed too. See the Status model in the Ad Campaigns tag.  **Live read, then write.** The ad set's switch is read from the platform first. When that live read shows it already in the requested state nothing is written (`updated: 0`, `skipped: 1`, with the reason). Otherwise the switch is written (`updated: 1`), read back and stored, and the delivery status of its ads (up to 20) is re-read and stored, so an immediate GET returns what the platform now reports. A stored switch never skips a write, and when the platform cannot be read the write always goes out. On Meta the check reads the ad set's own `status`, so a repeated request skips even while `platformAdSetStatus` reads `CAMPAIGN_PAUSED`, `WITH_ISSUES` or `IN_PROCESS`. 
 
 ### Examples
 
@@ -4345,7 +4345,7 @@ end
 
 Pause or resume a single ad
 
-Ad-scoped pause/resume: flips ONLY this ad's own switch (Meta `configured_status`, TikTok `operation_status`, Google ad group ad status, LinkedIn creative, Pinterest ad), never its parent ad set or campaign, so sibling ads keep running. X is the exception: its smallest switch is the line item. Thin wrapper over the `status` field of PUT /v1/ads/{adId}, for callers that want a URL symmetric to /v1/ads/campaigns/{campaignId}/status and /v1/ads/ad-sets/{adSetId}/status.  The ad's own switch is independent of its delivery status. An ad paused only because its campaign or ad set is off (`status: paused`, `configuredStatus: ACTIVE`) can still be switched off here, and switching an ad on under a paused campaign leaves it `paused` until the campaign is resumed. After the write the switch is read back from the platform and returned as `configuredStatus`, together with the resulting delivery `status`.  `{adId}` accepts the same identifier dialects as GET/PUT /v1/ads/{adId} (Zernio hex `_id`, Meta numeric `platformAdId`, or the creative's effective story/media IDs). `platform` is inferred from the ad, so it's not required in the body. Ads in terminal statuses (rejected, completed, cancelled) are skipped, and so is a request whose target already matches the ad's own switch as read LIVE from the platform (a stored value is never trusted alone, since the switch may have been changed in the platform's own UI). A skip also returns that live read. The rolled-up `status` never decides a skip. 
+Pauses or resumes a single ad without touching its ad set, campaign or sibling ads.  Ad-scoped pause/resume: flips ONLY this ad's own switch (Meta `configured_status`, TikTok `operation_status`, Google ad group ad status, LinkedIn creative, Pinterest ad), never its parent ad set or campaign, so sibling ads keep running. X is the exception: its smallest switch is the line item. Thin wrapper over the `status` field of PUT /v1/ads/{adId}, for callers that want a URL symmetric to /v1/ads/campaigns/{campaignId}/status and /v1/ads/ad-sets/{adSetId}/status.  The ad's own switch is independent of its delivery status. An ad paused only because its campaign or ad set is off (`status: paused`, `configuredStatus: ACTIVE`) can still be switched off here, and switching an ad on under a paused campaign leaves it `paused` until the campaign is resumed. After the write the switch is read back from the platform and returned as `configuredStatus`, together with the resulting delivery `status`.  `{adId}` accepts the same identifier dialects as GET/PUT /v1/ads/{adId} (Zernio hex `_id`, Meta numeric `platformAdId`, or the creative's effective story/media IDs). `platform` is inferred from the ad, so it's not required in the body. Ads in terminal statuses (rejected, completed, cancelled) are skipped, and so is a request whose target already matches the ad's own switch as read LIVE from the platform (a stored value is never trusted alone, since the switch may have been changed in the platform's own UI). A skip also returns that live read. The rolled-up `status` never decides a skip. 
 
 ### Examples
 
@@ -4416,7 +4416,7 @@ end
 
 Update portfolio bid strategy
 
-Renames or retargets a portfolio bid strategy. The strategy's status is output only on Google's side, so it cannot be changed here; remove a strategy in Google Ads. `type` is only needed alongside `targetCpa`/`targetRoas` to disambiguate the field Google writes to (TARGET_CPA and MAXIMIZE_CONVERSIONS both take a target CPA; TARGET_ROAS and MAXIMIZE_CONVERSION_VALUE both take a target ROAS); the strategy's family is otherwise immutable once created.
+Renames a shared Google Ads bid strategy or changes its target CPA or ROAS.  The strategy's status is output only on Google's side, so it cannot be changed here; remove a strategy in Google Ads. `type` is only needed alongside `targetCpa`/`targetRoas` to disambiguate the field Google writes to (TARGET_CPA and MAXIMIZE_CONVERSIONS both take a target CPA; TARGET_ROAS and MAXIMIZE_CONVERSION_VALUE both take a target ROAS); the strategy's family is otherwise immutable once created. 
 
 ### Examples
 
@@ -4487,7 +4487,7 @@ end
 
 Replace a campaign's ad schedule (dayparting)
 
-Replaces the campaign's whole ad schedule with the windows you send. This is a REPLACE, not a merge: windows you leave out stop serving.  Send `schedule: []` to clear dayparting, which returns the campaign to serving around the clock.  Google rules enforced here, so you get a named field instead of a criterion error: at most 6 windows per day, a window must end after it starts, windows on the same day may not overlap, `endHour` 24 is midnight and cannot carry minutes, and minutes are quarter-hours only (0, 15, 30, 45). `bidModifier` is 0.1-10.0; Google's 0 means \"off\" for devices only, so a window is switched off by leaving it out.  Windows are half-open (Google is exclusive of the end minute), so 09:00-12:00 and 12:00-17:00 on the same day are adjacent and both valid.  Google cannot edit an ad schedule in place (every AdScheduleInfo field is prohibited on update), so this removes the live criteria and creates the new ones in a single atomic mutate. The response is read back from Google and carries the new criterion ids. 
+Replaces the days and hours a Google campaign runs, or clears them so it serves around the clock.  This is a REPLACE, not a merge: windows you leave out stop serving.  Send `schedule: []` to clear dayparting, which returns the campaign to serving around the clock.  Google rules enforced here, so you get a named field instead of a criterion error: at most 6 windows per day, a window must end after it starts, windows on the same day may not overlap, `endHour` 24 is midnight and cannot carry minutes, and minutes are quarter-hours only (0, 15, 30, 45). `bidModifier` is 0.1-10.0; Google's 0 means \"off\" for devices only, so a window is switched off by leaving it out.  Windows are half-open (Google is exclusive of the end minute), so 09:00-12:00 and 12:00-17:00 on the same day are adjacent and both valid.  Google cannot edit an ad schedule in place (every AdScheduleInfo field is prohibited on update), so this removes the live criteria and creates the new ones in a single atomic mutate. The response is read back from Google and carries the new criterion ids. 
 
 ### Examples
 
@@ -4558,7 +4558,7 @@ end
 
 Update campaign assets
 
-Edits existing Google assets in place. Send updates with assetResourceName and the fields to change. An asset is shared: changes affect every attachment using it. Omitted fields stay unchanged. The operation consumes the Google operations budget and invalidates affected cached lists.
+Edits a Google campaign's existing assets, such as sitelinks or callouts, in place; the change applies everywhere that asset is used.  Send updates with assetResourceName and the fields to change. An asset is shared: changes affect every attachment using it. Omitted fields stay unchanged. The operation consumes the Google operations budget and invalidates affected cached lists. 
 
 ### Examples
 
@@ -4629,7 +4629,7 @@ end
 
 Update campaign conversion goals
 
-Sets `biddable` on campaign goals, switches `goalConfigLevel`, and/or points the campaign at a custom conversion goal, in one mutate. `customConversionGoalId: null` clears it; Google refuses that (400) while the campaign stays at CAMPAIGN level with no biddable goals, so send `goalConfigLevel: CUSTOMER` with it to fall back to the account goals. Returns the re-read campaign goals.
+Changes which conversions a Google campaign bids on, switching between account-default goals, its own goals or a custom goal.  Sets `biddable` on campaign goals, switches `goalConfigLevel`, and/or points the campaign at a custom conversion goal, in one mutate. `customConversionGoalId: null` clears it; Google refuses that (400) while the campaign stays at CAMPAIGN level with no biddable goals, so send `goalConfigLevel: CUSTOMER` with it to fall back to the account goals. Returns the re-read campaign goals.
 
 ### Examples
 
@@ -4700,7 +4700,7 @@ end
 
 Edit a Google campaign's device, location, excluded location, or language targeting
 
-Google Ads compliance row M.10: geo and language targeting set at creation must stay editable afterwards. Send at least one of `devices`, `locations`, `excludedLocations`, `languages`, `locationTargetingType`; each provided field REPLACES that field's existing criteria on the campaign (a full set, not a delta). Fields left out of the body are untouched. Google only; every other platform returns 501.  `devices` is the full set of device bid modifiers: a supported device you leave out is switched off with a bid modifier of 0, since Google cannot remove a device criterion. A device the campaign's channel does not carry, and a set that switches every device off, both return 422.  `locations` accepts the same shapes as campaign creation: a bare array of ISO country codes, or an object with `countries`/`regions`/`cities`/`zips`/`metros` key lists (`key` from GET /v1/ads/targeting/search?dimension=geo). Excluded locations are left untouched by `locations`. An empty location list returns 400 instead of removing every criterion: a Google campaign with no location criteria targets every country, so omit `locations` to leave targeting alone.  `excludedLocations` takes the same two shapes and replaces the campaign's negative location criteria (the places it never serves in), leaving the targeted `locations` untouched. An empty list (or `{}`) removes every exclusion. Radius exclusions are not supported. A place cannot be both targeted and excluded: a request whose result would leave one on both sides returns 400 before anything is written, and moving a place from one side to the other in the same request is applied atomically. Example: `{ \"platform\": \"google\", \"targeting\": { \"excludedLocations\": { \"countries\": [\"CA\"], \"regions\": [\"21137\"] } } }`.  The removes and the creates go out in ONE Google `googleAds:mutate`, so a failed edit leaves the campaign's previous set intact rather than a half-applied one.  `languages` is an array of Google's language codes (ISO 639-1, plus variants such as `zh_CN`); an unknown code returns 400.  `locationTargetingType` switches who the location targeting reaches: `presence` (people in or regularly in the locations) or `presence_or_interest` (also people searching for or interested in them). Example: `{ \"platform\": \"google\", \"targeting\": { \"locationTargetingType\": \"presence\" } }`.  The response includes the refreshed `devices`/`locations`/`excludedLocations`/`languages` state read back from Google after the edit, and invalidates the cached copy `GET` on this campaign would otherwise keep serving.  **Demand Gen:** Google keeps a Demand Gen campaign's locations and languages on its ad groups and refuses them on the campaign. When the campaign has one ad group they are written there and the response carries its `adGroupId` (the campaign-level `locations`/`languages` read back then stay empty). With several ad groups the call returns 400 naming them: edit each one with PUT /v1/ads/{adId} `targeting` on an ad of that ad group. Campaigns migrated from Discovery that still target on the campaign keep being written there. `devices` and `locationTargetingType` stay campaign-level. `excludedLocations` is not available on Demand Gen yet and returns 400. 
+Changes a Google campaign's device, location, excluded location or language targeting after it was created.  Google Ads compliance row M.10: geo and language targeting set at creation must stay editable afterwards. Send at least one of `devices`, `locations`, `excludedLocations`, `languages`, `locationTargetingType`; each provided field REPLACES that field's existing criteria on the campaign (a full set, not a delta). Fields left out of the body are untouched. Google only; every other platform returns 501.  `devices` is the full set of device bid modifiers: a supported device you leave out is switched off with a bid modifier of 0, since Google cannot remove a device criterion. A device the campaign's channel does not carry, and a set that switches every device off, both return 422.  `locations` accepts the same shapes as campaign creation: a bare array of ISO country codes, or an object with `countries`/`regions`/`cities`/`zips`/`metros` key lists (`key` from GET /v1/ads/targeting/search?dimension=geo). Excluded locations are left untouched by `locations`. An empty location list returns 400 instead of removing every criterion: a Google campaign with no location criteria targets every country, so omit `locations` to leave targeting alone.  `excludedLocations` takes the same two shapes and replaces the campaign's negative location criteria (the places it never serves in), leaving the targeted `locations` untouched. An empty list (or `{}`) removes every exclusion. Radius exclusions are not supported. A place cannot be both targeted and excluded: a request whose result would leave one on both sides returns 400 before anything is written, and moving a place from one side to the other in the same request is applied atomically. Example: `{ \"platform\": \"google\", \"targeting\": { \"excludedLocations\": { \"countries\": [\"CA\"], \"regions\": [\"21137\"] } } }`.  The removes and the creates go out in ONE Google `googleAds:mutate`, so a failed edit leaves the campaign's previous set intact rather than a half-applied one.  `languages` is an array of Google's language codes (ISO 639-1, plus variants such as `zh_CN`); an unknown code returns 400.  `locationTargetingType` switches who the location targeting reaches: `presence` (people in or regularly in the locations) or `presence_or_interest` (also people searching for or interested in them). Example: `{ \"platform\": \"google\", \"targeting\": { \"locationTargetingType\": \"presence\" } }`.  The response includes the refreshed `devices`/`locations`/`excludedLocations`/`languages` state read back from Google after the edit, and invalidates the cached copy `GET` on this campaign would otherwise keep serving.  **Demand Gen:** Google keeps a Demand Gen campaign's locations and languages on its ad groups and refuses them on the campaign. When the campaign has one ad group they are written there and the response carries its `adGroupId` (the campaign-level `locations`/`languages` read back then stay empty). With several ad groups the call returns 400 naming them: edit each one with PUT /v1/ads/{adId} `targeting` on an ad of that ad group. Campaigns migrated from Discovery that still target on the campaign keep being written there. `devices` and `locationTargetingType` stay campaign-level. `excludedLocations` is not available on Demand Gen yet and returns 400. 
 
 ### Examples
 
@@ -4771,7 +4771,7 @@ end
 
 Update a Performance Max asset group
 
-Change the name, status (ENABLED or PAUSED), final URLs or display paths. Only the fields sent are written; null on path1 or path2 clears it. Change assets with the /assets endpoint and product targeting with /listing-group-filters. validateOnly: true validates without writing.
+Changes a Performance Max asset group's name, status, final URLs or display paths.  Change the name, status (ENABLED or PAUSED), final URLs or display paths. Only the fields sent are written; null on path1 or path2 clears it. Change assets with the /assets endpoint and product targeting with /listing-group-filters. validateOnly: true validates without writing. 
 
 ### Examples