b2c-app-builder 0.221.39

package/catalog/generated/firstparty/knowledge/growth/non-competitor-angle-discovery.md

@@ -0,0 +1,80 @@
+# Non-Competitor Angle Discovery
+
+Use this before the first paid creative batch.
+Direct competitor ads are a trap when every angle is a logo swap of the same category tropes.
+Fresh angles come from four hunt sources, then from the same customer with a new motivation.
+
+Load `paid-user-acquisition.md` for channel, cadence, and kill rules.
+Load `../design/design-worthiness.md` for the logo-swap originality rule.
+Load `../design/audience-derived-identity.md` for the same test on product identity.
+
+## Contents
+
+- 1. Four Hunt Sources
+- 2. Same Customer, New Motivation
+- 3. Competitor-Wrong Gate
+- 4. Ethics And Vendors
+- 5. Outputs
+- 6. Common Failure Modes
+
+## 1. Four Hunt Sources
+
+Record a Non-Competitor Angle Hunt table in `growth/PAID_UA.md`.
+
+Hunt these source kinds before spend:
+
+- print — slogans, layouts, and frames from print, not from in-category ads
+- cross-niche — a psychological trigger from a different market
+- ad-library-adjacent — public ad libraries for the same targeting or an adjacent or international category, not the direct incumbent
+- second-profile — a second motivation for the same ideal customer
+
+The hunt must name all four kinds.
+Not every hunted angle must ship.
+At least one adopted angle must come from a non-competitor source.
+
+Do not start with a paid bookmark tool when these four sources can run.
+
+## 2. Same Customer, New Motivation
+
+When the primary profile saturates, expand motivations for the same customer.
+Do not only raise budget.
+Do not switch to a new customer until the current customer has a second truthful motivation.
+
+Record that expansion in the hunt table.
+The product must still serve the new motivation.
+
+## 3. Competitor-Wrong Gate
+
+Each adopted angle records a competitor-wrong result:
+
+- `original` — a logo swap onto a category competitor would not still work
+- `logo-swap` — the angle still works if you only change the wordmark
+
+At least one adopted angle must be `original`.
+A batch of only `logo-swap` angles is not ready for first spend.
+
+This is the paid-creative bind of design-worthiness originality.
+Do not invent a beauty score.
+
+## 4. Ethics And Vendors
+
+Adapt the trigger.
+Never clone competitor assets, footage, or claims.
+Never add fake scarcity.
+
+Foreplay and similar bookmark tools are optional.
+They are not a B2C App Builder default.
+
+## 5. Outputs
+
+- Non-Competitor Angle Hunt table in `growth/PAID_UA.md`
+- angle IDs in `CONTENT_ASSETS.md` that cite hunt rows
+- one adopted `original` angle before first spend
+- same-ICP new-motivation row after primary saturation
+
+## 6. Common Failure Modes
+
+- First creative batch mined only from direct competitor ads
+- Raising budget when the primary motivation is exhausted
+- Adopted angles that all pass a logo-swap onto a category rival
+- Cloning competitor claims or creatives

package/catalog/generated/firstparty/knowledge/growth/paid-user-acquisition.md

@@ -0,0 +1,283 @@
+# Paid User Acquisition
+
+Use this before planning paid ads, Apple Search Ads, Meta/TikTok/Google campaigns, web-to-app paid traffic, custom product page campaign routing, paid creative tests, MMP/ad-network SDK choices, or any launch claim that the app is ready to spend on acquisition.
+
+This reference factors in RevenueCat's May 2026 article, "How to build a UA system when you're a one-person team": https://www.revenuecat.com/blog/growth/how-to-build-a-ua-system-when-youre-a-one-person-team. Treat it as a practical solo-founder operating model, not universal law. The transferable shape is small, complete, measurable, and cadence-driven: one channel, creative production, tracking, and schedule.
+
+Paid spend is a founder-only gate. Agents may prepare the system, docs, creative plan, dashboards, and stop/scale rules, but must not launch campaigns, connect ad accounts, change budgets, create paid trials, or commit spend without approval.
+
+Load [`paid-tool-routing.md`](../operations/paid-tool-routing.md) before replacing paid ASO/MMP/ad tools, ad-network accounts, RevenueCat paid features, PostHog paid features, or App Store/Play/Google/Meta/TikTok account data with free or manual reporting. Load `analytics-attribution.md`, `revenue-monetization.md`, and `app-store-listing-prep.md` when paid UA touches tracking, LTV/CPA, store conversion, screenshots, or ad creative. Update `CONTENT_ASSETS.md`; load `remotion-content-assets.md` when Remotion-rendered ad creative or repeatable variants are involved.
+
+Load `localization-market-research.md` before choosing Apple Search Ads or paid storefronts/geos: the target countries come _from_ the `LOCALIZATION_MARKET_RESEARCH.md` priority tiers (keyword popularity, difficulty, demand per storefront), not from wherever the app happens to be available. Spending into a market with no demand evidence or no localized paywall is the same translate-first mistake in paid form.
+
+## Contents
+
+- Fit Gate
+- One-Channel Choice
+- Creative Production System
+- Tracking And Baseline
+- Blended Report
+- Weekly Schedule
+- Stop And Scale Rules
+- Draft Status And Kill Window
+- Founder-Only Gates
+- Outputs
+- Common Failure Modes
+
+## Fit Gate
+
+Paid UA is plausible when:
+
+- the app has a real store/beta/web destination that can accept traffic
+- onboarding and paywall flow can convert paid visitors without contradicting store, privacy, or subscription copy
+- RevenueCat or the approved subscription source can show LTV, trial starts, cohorts, and entitlement truth
+- analytics can distinguish landing/store clicks, installs or opens, onboarding, paywall reach, purchase/trial, entitlement activation, and retention
+- the product has at least 3-5 truthful creative angles tied to the 11-star accepted product experience
+- the founder has approved a test budget and account access path
+
+Defer or reject paid UA when:
+
+- store pages, privacy/terms, support, pricing, or purchase validation are not ready
+- no baseline period exists for RevenueCat, App Store Connect, PostHog, and self-reported attribution
+- the launch cannot absorb traffic because onboarding, paywall, restore, support, or crash/error monitoring is weak
+- the only plan is "try ads" without a channel, creative cadence, target event, dashboard, and stop rule
+- paid spend or ad-account access is unapproved
+
+Record the fit decision in `PAID_UA.md`, update `state/business-state.json` `lanes.paid_user_acquisition`, and trace the decision into `state/LAUNCH_TRACE.md`.
+
+## One-Channel Choice
+
+Start with one paid channel. Do not launch Meta, TikTok, Google, and Apple Ads at the same time unless the founder has a team or explicit reason to absorb separate creative, reporting, account-management, and attribution costs.
+
+Default channel choice:
+
+- **Meta Ads**: default first channel for many subscription apps when the ICP is broad enough and the product can produce varied visual/pain-point creatives.
+- **TikTok**: consider when the buyer is younger, the product has strong native video hooks, and the launch can refresh creatives faster.
+- **Google web-to-app**: consider when search/demand capture or web funnels matter and the app can preserve campaign context into checkout or install.
+- **Apple Ads**: consider when search intent is clear, keyword economics are plausible, and App Store listing/custom product pages are ready.
+
+Sequencing against creator sponsorships: hand-negotiated creator deals (`influencer-sponsorship-engine.md`) are typically the higher-multiple channel at small budgets, while ad platforms return lower multiples but scale without per-deal effort. When sponsorships have proven the demo moment and payback economics, their winning creative and creators become the seed for the first paid channel — including paying one proven creator to make ads directly rather than posting to their own page.
+
+Document:
+
+- selected channel and why
+- rejected channels and why now is too early
+- target event: usually trial, direct subscription, paywall-qualified event, or another high-intent proxy; avoid install-only campaigns as the main growth pillar unless the launch explicitly uses them for creative testing
+- campaign destination: store listing, custom product page, web funnel, landing page, or web purchase link
+- founder-approved budget cap and date range, or blocker if not approved
+
+When the campaign destination is a custom product page, load [`surfaces-b2c.md`](../design/surfaces-b2c.md) for Apple's CPP/PPO/In-App Event numeric ceilings and [`design-room.md`](../design/design-room.md) before committing spend. Model the CPP as a real surface in the screen map, update `DESIGN.md`, render the Design Room, and retain the revision in Git. It is not a campaign-only asset.
+
+## Creative Production System
+
+Before spend, define the first creative batch:
+
+- before drafting ad headlines or primary/body text, load `knowledge/words/no-slop-writing.md` — its §7 covers ad headline/primary-text limits; verify current platform character limits via the `ad-creative` skill or platform docs, and lead with the concrete benefit or hook, not a throat-clearing opener
+- 3-5 new assets per week for an early solo cadence
+- one ad group or equivalent simple structure for the first test
+- angle map: pain, aspiration, identity, objection, comparison, proof, result, or use case
+- product visibility: real app UI, store screenshot, app-flow recording, Remotion render, or approved generated/supporting asset
+- copy and claim constraints from `APP_STORE_LISTING.md`, `trust/PRIVACY.md`, `trust/TERMS.md`, `revenue/REVENUE_OPS.md`, and `trust/SECURITY.md`
+- each creative tied to `11_STAR_EXPERIENCE.md`, `CONTENT_ASSETS.md`, and `state/LAUNCH_TRACE.md`
+
+Use `remotion-content-assets.md` when repeatable ad variants, motion screenshots, captions, cutdowns, or local render proof are needed. Use Higgsfield only after design-system constraints and paid-tool routing are recorded. Use real app UI wherever a user must inspect the actual product. Before any creative is scored or spent on, run it through [`quality-lens.md`](../design/quality-lens.md)'s Anti-Generic Checks — design-token compliance and a `brain_activity` virality score both pass a generic, off-brand, or overclaiming asset; taste is the check neither of them performs.
+
+**Competitor ad-library mining.** Before inventing angles, mine the platform's public ad library (Meta Ad Library for Meta/Instagram) for the incumbent and adjacent apps: filter for ads with high impressions and long run time — nobody keeps paying for an ad that does not return — and for each winner ask why it works and how the structure maps to this app's truthful claims. Replicate the angle and structure with this app's real UI and claims, never the competitor's assets, footage, or claims verbatim. Authentic low-production formats frequently outperform polished ones; test them before assuming production value is the constraint. Record mined angles in the angle map above.
+
+Load [`non-competitor-angle-discovery.md`](non-competitor-angle-discovery.md) before the first paid creative batch. Direct competitor ads are not enough. Hunt print, cross-niche, Ad Library adjacent, and second-profile sources. Expand the same ICP with a new motivation after the primary profile saturates. At least one adopted angle must be original under a competitor-wrong / logo-swap test.
+
+### Higgsfield Creative Production
+
+Before generating any Higgsfield asset: confirm spend with the founder per `paid-tool-routing.md`, carry `DESIGN.md` tokens (palette, type mood, shapes, texture, motion energy, banned aesthetics, surface) in every prompt, and record every output in `CONTENT_ASSETS.md`.
+
+**Routed recipes** (bodies in `tool-recipes/visual-and-motion-production.md` — do not re-copy here):
+
+- **App Store URL → UGC Ad Batch (Click-to-Ad)** — webproducts fetch → avatar pick → parallel mode batch → score → save winners → founder approval.
+- **Virality Closed Loop** — generate → score → decision rules → iterate → record → paid distribution only after a score is recorded.
+- **Cheap-First Direction** — z_image drafts → pick direction → production model → founder selects. Apply only when presented as a spend-reduction option at the spend-confirmation prompt; never silently.
+
+**DTC static-ad batch path.** Browse available formats first (`higgsfield marketing-studio ad-formats list`), then:
+
+```
+higgsfield marketing-studio dtc-ads generate --format-id <id> [--brand-kit-id <id>] [--batch-size 1..20]
+```
+
+`--format-id` is mandatory. `--brand-kit-id` requires a prior paid brand-kit fetch — confirm spend before running.
+
+**Hook A/B protocol.** Before committing to a production run, test breadth before depth: run 4 hooks × 1 mode first to find the winning hook, then 1 hook × 4 modes to find the winning mode. Score each variant with `brain_activity` before expanding.
+
+**Click-to-Ad `--url` note.** The `--url <app-store-url>` shortcut bypasses brief injection — `DESIGN.md` tokens are NOT carried automatically. Always inject them explicitly into `--prompt`.
+
+**Win-back / re-engagement creatives.** For lapsed-user campaigns, generate `marketing_studio_video --mode product_review` ads with "do you miss X" hooks, plus `product-photoshoot --mode lifestyle_scene` static re-engagement visuals; score the video with `brain_activity` before allocating any spend. Record `virality_score` and `hook_dmn_risk` in `CONTENT_ASSETS.md` before distributing.
+
+## Tracking And Baseline
+
+Finish the tracking baseline before the first paid dollar. Record this as **MMP Before Spend** in `growth/PAID_UA.md`.
+
+Load `appsflyer-mmp.md` when paid UA is in scope.
+
+AppsFlyer is the default MMP on that path.
+Set `$appsflyerId` on RevenueCat before trial or purchase.
+A founder MMP waiver is required for any other stack.
+A founder MMP waiver must be a dated table with founder identity, named stack, and approval state approved.
+`check:paid-ua` fails a spend-ready lane that lacks the bridge and has no waiver.
+
+Do not treat a network SDK plus RevenueCat plus a blended sheet as spend-ready by itself.
+
+Decide the rest of the tracking stack from current product phase:
+
+- RevenueCat for LTV, cohorts, subscription events, trial starts, purchase state, and entitlement truth
+- selected ad-network SDK or native reporting when the channel requires it
+- App Store Connect or Google Play Console for store/source signals
+- PostHog for product analytics, identity, attribution, funnels, and dashboards
+- a simple spreadsheet or `growth/paid-ua-report.csv` for the blended weekly report
+
+Create a baseline before campaign launch:
+
+- RevenueCat daily trials, purchases, revenue, cohorts, and LTV
+- App Store Connect or Play installs/product page views where available
+- PostHog events from app open through onboarding, paywall, purchase, entitlement, activation, and retention
+- self-reported attribution answers from onboarding/signup/waitlist
+- current organic/social traffic notes so uplift is not confused with paid performance
+
+For iOS, document that ad-platform attribution can under-report or over-report. Use baseline uplift across RevenueCat, ASC, PostHog, and self-reported attribution instead of pretending attribution is perfect.
+
+## Blended Report
+
+Create a weekly report that lets the founder decide whether to continue, pause, or scale. When the paid UA lane is done, `check:paid-ua` requires the report file to actually look like one — a header naming spend plus downstream economics (CPA/LTV/revenue/ROAS/payback) and at least one data row; an empty or stub file fails the gate.
+
+Minimum columns:
+
+- date
+- channel
+- campaign
+- spend
+- impressions
+- clicks
+- installs or app opens
+- onboarding starts
+- paywall views
+- trial starts
+- purchases
+- entitlement active count
+- revenue
+- RevenueCat LTV window
+- CPA or cost per trial/direct subscription
+- blended ROAS or projected payback
+- winning angle
+- virality_score (brain_activity overall)
+- hook_dmn_risk (Default Mode risk from brain_activity)
+- notes and next action
+
+The report should compare paid traffic against organic baseline and category benchmarks where available. It should be good enough to catch product-side drop-offs, not only media-buying changes.
+
+## Weekly Schedule
+
+Use a calm twice-weekly operating cadence for low-budget tests instead of reactive daily edits:
+
+- Monday: refresh campaign and blended report, compare against previous week, decide pause/keep/scale candidates, and choose one paywall/onboarding test hypothesis.
+- Tuesday: plan and produce the next 3-5 creative assets from winning or unresolved angles.
+- Wednesday: check early creative delivery, pause older non-spending or clearly weak assets only when there is enough signal, and upload 1-2 replacements.
+- Thursday: quick anomaly and report check.
+- Friday: upload remaining assets when needed, then decide whether to scale, hold, reduce, or pause based on the full week.
+- Daily: spend 15 minutes checking pacing and major anomalies only.
+
+Automated ad-platform rules may be planned when supported, but live setup or budget changes require founder approval.
+
+## Stop And Scale Rules
+
+Every rule below evaluates against a number recorded in `growth/PAID_UA.md` under **Decision Thresholds** — an agent (or the founder at 11pm) must be able to read the week's report and know which side of each line it landed on. The defaults are starting points to adjust per category evidence, the same way the crash-free threshold works in `post-launch-operations.md`; changing a threshold is a recorded edit with a reason, never a silent re-read:
+
+- **Attribution tolerance** — default ±20%. The maximum disagreement between RevenueCat, ASC/Play console, PostHog, and self-reported attribution before the numbers stop being readable.
+- **Payback window** — default 90 days. Blended CPA must be recoverable from realized LTV (Economics Snapshot in `revenue/REVENUE_OPS.md`) inside this window.
+- **Creative signal floor** — default 2x target CPA in spend, or 7 days, per creative — whichever comes first — before a creative may be judged or killed.
+- **Scale trigger** — default 14 consecutive days at or under target CPA at the founder-approved spend level before any budget increase is proposed.
+
+Stop, hold, or reposition when:
+
+- baseline cannot be established
+- spend is active but the attribution sources disagree beyond the recorded tolerance for a full reporting week
+- blended CPA exceeds what the recorded payback window supports against realized LTV, or trial conversion breaks the payback math
+- paywall reach, purchase conversion, entitlement activation, or retention drops versus organic traffic
+- creatives are being killed before reaching the signal floor — churning creative faster than the budget can generate readable signal
+- the only positive metric is impressions, clicks, or installs
+
+Scale only after:
+
+- the scale trigger has held — target CPA or better, at the approved spend level, for the recorded number of consecutive days
+- one or more creative angles repeat with downstream revenue or activation evidence
+- weekly creative production can keep up without lowering quality
+- onboarding/paywall/revenue experiments and support paths are stable
+- the founder approves higher spend or new channels
+
+Outsource only after identifying the bottleneck:
+
+- creative producer when asset production is slowing learning
+- data analyst when reporting quality is the bottleneck
+- part-time UA manager when campaign operations consume too much founder time
+
+Outsource execution, not strategy, until the founder understands the channel economics.
+
+## Draft Status And Kill Window
+
+Create campaigns as **PAUSED** drafts until the founder approves live spend. An agent may prepare the draft. An agent must not enable delivery.
+
+Record a **48-hour kill window** in `growth/PAID_UA.md`. Review CPA, payback, and product drop-off at 48 hours after first spend. Pause the campaign when the kill rule fires. Do not wait for the next weekly review to stop a broken test.
+
+Load [`ios-consumer-playbook-intake.md`](./ios-consumer-playbook-intake.md) before copying an external iOS paid-UA tutorial into this packet.
+
+## Founder-Only Gates
+
+Always ask before:
+
+- connecting ad accounts or creating campaigns
+- unpausing a PAUSED draft or enabling delivery
+- installing or enabling ad-network SDKs that change privacy disclosures or ATT posture — load [`privacy-terms.md`](../trust/privacy-terms.md) for the policy/questionnaire update and [`store-console-workflow.md`](../store/store-console-workflow.md) for Data safety/privacy label changes before enabling one
+- committing paid spend, changing budgets, or turning on automated rules that affect spend
+- using paid MMP/ad/ASO tooling
+- changing pricing, trials, intro offers, subscription terms, paywall behavior, or legal copy
+- launching custom product pages, web funnels, web purchase links, or public ad creative
+- sharing screenshots, creator/founder likeness, generated assets, or claims publicly
+
+## Outputs
+
+Create or update:
+
+- `PAID_UA.md`: fit gate, channel choice, creative system, tracking baseline, MMP Before Spend, PAUSED drafts, 48-hour kill window, blended report, schedule, stop/scale rules, founder-only gates, and traceability
+- `growth/paid-ua-report.csv` when spend is planned or active
+- `analytics/ANALYTICS.md` and `analytics/analytics-plan.html` with campaign, ad-network, baseline, and RevenueCat/ASC/PostHog dashboard definitions
+- `revenue/REVENUE_OPS.md` with LTV/CPA/payback assumptions and RevenueCat cohort source
+- `CONTENT_ASSETS.md` and optional Remotion/Higgsfield artifacts for repeatable creative production
+- `APP_STORE_LISTING.md`, `store/STORE_CONSOLE.md`, or custom product page notes when the campaign destination is a store surface
+- `state/LAUNCH_TRACE.md` rows connecting research, 11-star slice, channel, creative angles, paywall/revenue, analytics, privacy/legal, and proof
+
+Run:
+
+```bash
+npm run check:paid-ua -- --root .
+```
+
+## Common Failure Modes
+
+- Marking paid acquisition ready while ad spend is still unapproved.
+- Launching multiple channels before one channel has a creative/reporting loop.
+- Treating install volume as success while paywall reach, purchase, entitlement, and retention are unknown.
+- Adding a paid ad SDK without updating analytics, privacy, ATT, store disclosures, and `SECRETS.md`.
+- Using RevenueCat products and paywalls without checking whether CPA can fit LTV and trial conversion.
+- Scaling creative volume before the app has real product UI, clear claims, and repeatable angle evidence.
+- First creative batch mined only from direct competitor ads, with no original non-competitor angle.
+- Copying a big-team reporting stack without the AppsFlyer `$appsflyerId` bridge.
+- First spend with only a network SDK and a sheet, and no founder MMP waiver.
+- Shipping ad copy that reads like generic ad-speak (weasel attribution, importance puffery) instead of passing the `no-slop-writing.md` self-check.
+
+## Anti-Patterns (Spend And Sequencing Traps)
+
+The paid-acquisition entries in the monetization-and-growth digest in [`revenue-monetization.md`](../money/revenue-monetization.md) §3. These are the cozy moves that feel like progress but burn budget. The Meta dashboard being "right there and shiny" is not a reason to spend.
+
+**8. Run ads before you can read the result.** Below a meaningful daily conversion volume, every paywall, trial, and creative test is statistical noise — you spend the budget, get random numbers, declare them "insights," and build strategy on a coin flip. Do not start paid spend until the baseline window and minimum volume needed for a readable result are defined (see Tracking And Baseline and Stop And Scale Rules: "baseline cannot be established" is a stop condition). If volume is too low to reach significance, the correct move is more organic/ASO signal or a longer window, not more spend.
+
+**9. Optimize ad creative first.** The fun part — hooks, fonts, trending audio — is not the constraint. The order that works is **paywall → onboarding → ads**: paywall (and trial/price) LTV is the bottleneck, onboarding is the conversion ramp into it, and ads only multiply whatever those two already do. Pouring traffic into a leaky paywall is filling a bucket with no bottom and calling it marketing. The Fit Gate already requires `revenue-monetization.md` and `onboarding-conversion.md` to be in good shape before spend; do not invert the order by polishing creatives while the paywall is unproven.
+
+**11. Chase views / vanity reach.** A 500K-view screenshot for the group chat is a hug, not a business — views without buying intent do not convert (one founder turned broad emotional reach into $400). Optimize for installs → paywall reach → purchase → retention, not impressions. The Stop And Scale Rules already flag "the only positive metric is impressions, clicks, or installs"; the same trap on the organic side is covered in [`viral-growth-loops.md`](./viral-growth-loops.md) Common Failure Modes (optimizing for views instead of installs/paywall/purchases).
+
+- Calling paid UA "done" while `PAID_UA.md`, `analytics/ANALYTICS.md`, `revenue/REVENUE_OPS.md`, `CONTENT_ASSETS.md`, `APP_STORE_LISTING.md`, or `state/LAUNCH_TRACE.md` disagree.

package/catalog/generated/firstparty/knowledge/growth/ugc-creator-engine.md

@@ -0,0 +1,259 @@
+# UGC Creator Engine
+
+Use this when the launch depends on TikTok/Reels/Shorts, creator-led organic growth, founder-led content, UGC ads, or post-launch social distribution. This reference factors in the user-provided Sprout UGC Playbook Part 1 PDF, extracted as "Zero to One: How a UGC engine actually starts" by Aaron Paul, published May 2026.
+
+The Day 0 frame: do not build a mature creator machine before a format works. Run a 90-day founder-led format-discovery experiment with a tiny roster, high feedback intensity, and strict stop/scale rules.
+
+Load `viral-growth-loops.md` first when UGC depends on product-led referral/share mechanics, social-comment loops, paywall timing, creator codes, or share-to-unlock behavior. This file owns creator operations; `viral-growth-loops.md` owns the product/content/revenue/analytics system that makes attention compound.
+
+Load `influencer-sponsorship-engine.md` instead when the plan is paying creators who already have an audience to integrate the app into their own content — sponsorship outreach, CPM/view-guarantee deals, brand-account credibility, meme-page amplification. This file's creator ops assume the program owns new niche accounts and the algorithm is the distribution engine; sponsorships rent an existing audience and price on views, and the two lanes use different payment shapes and qualification filters.
+
+## Contents
+
+- Fit Gate
+- Day 0 Operating Model
+- Creator Sourcing
+- Onboarding And Contracts
+- Payment Shape
+- Format Discovery Loop
+- Tracking And Analytics
+- Design And Product Implications
+- Outputs
+- Common Failure Modes
+
+## Fit Gate
+
+UGC is a good candidate when the app has:
+
+- a visible product moment that can be shown in a phone screen, laptop screen, demo, or screenshot
+- an emotional hook: anxiety, frustration, desire, identity, aspiration, relief, embarrassment, pride
+- a large consumer or prosumer audience
+- a natural comment-bait tension or share/tag reason
+- a product name or visual identity that viewers can recognize from the video frame
+
+Deprioritize UGC when:
+
+- there is no visible product moment
+- the product is pre-product or has no real workflow to film
+- the audience is narrow, committee-led, or high-ticket enterprise
+- the category is regulated and claims/compliance cannot be reviewed
+- the product cannot tolerate rough, fast, creator-native presentation
+
+Record the fit decision in `growth/UGC_PLAYBOOK.md` and map it to `state/LAUNCH_TRACE.md` and `11_STAR_EXPERIENCE.md`. A creator hook should be a tiny, truthful version of the product's complete product experience, not a separate marketing fantasy.
+
+## Day 0 Operating Model
+
+Run Day 0 as a format-discovery experiment:
+
+- commit to 90 days before judging the channel
+- use 3-5 creators, not a large roster
+- expect roughly 3-4 approved videos per creator per week
+- plan for 9-12 approved videos per week across a 3-creator roster
+- write every script for the first 4-8 weeks
+- approve every early cut before posting
+- give same-day, specific, time-coded feedback
+- give each format 5-8 reps before judging it
+- look for a format that hits 2-3 times, ideally across 2+ creators, before scaling
+
+Do not hire a large agency, buy production gear, build a heavy attribution stack, or run a 20-creator roster before the first repeatable format appears.
+
+## Creator Sourcing
+
+The first search target:
+
+- personal TikTok accounts with roughly 500-5,000 followers
+- US, UK, Canada, Australia, or whichever geography matches the paying buyer and pricing
+- camera-native, clear speakers, reliable responders, coachable, comfortable with feedback
+- not necessarily niche experts; the founder trains the niche through scripts
+
+Day 0 sourcing workflow:
+
+1. Open a founder/brand creator account and post 5-10 niche videos before recruiting, if feasible.
+2. Scout roughly 100 creators on TikTok or an approved creator marketplace.
+3. Use paid-tool routing before using Sideshift or another paid creator platform.
+4. Keep the first DM short, lead with paid collaboration, explain the app in one concrete line, and ask for email/phone rather than closing in DMs.
+5. Expect a small percentage of cold DMs to become onboarded creators.
+
+Creator-owned personal audience is not the point. The creator runs a new niche-themed account for the program. The algorithm, not their existing follower base, is the distribution engine.
+
+## Onboarding And Contracts
+
+Compress week 1:
+
+- Day 1: sign a short contractor agreement before account credentials change hands
+- Day 2: creator creates the new niche-themed brand account on their own device and shares login as backup
+- Day 3: founder sends scripts in the simplest shared doc/text thread that works
+- Days 4-7: creator films, founder reviews, creator revises, then posts
+- end of week 1: quick async check-in and first iteration notes
+
+Contract checklist for Day 0:
+
+- brand/account ownership and handover
+- content usage and paid-ad reuse rights
+- base pay, bonus trigger, view source, and measurement window
+- termination terms and in-flight bonus handling
+- tax forms and payout method
+- disclosure requirements: commercial relationship, partner language, hashtags, and platform rules
+
+Get legal review before scale. Do not leave account ownership, ad usage, or tax paperwork ambiguous.
+
+## Payment Shape
+
+The Day 0 payment structure should keep creators engaged while the program has no proven hit rate:
+
+- base pay per approved video
+- performance bonus after a meaningful view threshold
+- no pure-performance-only structure at Day 0
+- no affiliate or rev-share-first structure before a repeatable format exists
+
+Sprout PDF benchmark to adapt, not blindly copy:
+
+- base: roughly $30-40 per approved video
+- bonus trigger: meaningful breakout at around 50K views
+- rough flat bonus after the trigger, then tune to category economics
+- later scale can use declining CPM tiers so mega-hits do not destroy unit economics
+
+Budget planning:
+
+- floor: base-only spend for 3 creators over 13 weeks
+- working case: base plus mid-hit bonuses
+- breakout case: monthly spend can jump sharply, which can be a success signal if unit economics work
+
+Any real creator spend, paid platform spend, legal spend, or payout commitment is founder-approved.
+
+## Format Discovery Loop
+
+Test:
+
+- hooks: first 1-3 seconds, usually the highest-leverage variable
+- formats: talking-head, demo, POV, skit, fake interview, street, duet, teardown, before/after, challenge
+- angles: pain, aspiration, identity, objection, comparison, result, use case
+- characters: self, expert, beginner, skeptic, exaggerated persona, antagonist, customer, friend
+- CTA mechanic: comment bait, tag/share, profile search, waitlist, app search, store link, demo request
+
+Define a format as a repeatable structure, not a topic. A format includes hook shape, beat order, product insertion, tension, and CTA mechanic. Steal the skeleton from proven patterns, not the skin, lines, scenes, or copyrighted material.
+
+Before writing scripts for `ugc/script-bank.md`, load `knowledge/words/no-slop-writing.md` for banned words/patterns and weasel-attribution claims — but the voice being protected here is the creator's own natural spoken voice, not this skill's tone or the app's `strategy/BRAND.md`. A script that reads like brand copy instead of a real person talking on camera fails before any slop check does.
+
+**Script bank and judge panel.** A script earns filming or generation spend by surviving a judge panel. It does not earn spend by reading well to its author. Run the panel as separate reviewer passes with fresh context (subagents, or the council mechanism). Give each judge exactly one job: pacing, vocabulary, idea strength, or structure. Ground at least one judge entirely in a real creator's transcript corpus. Pick an imperfect source on purpose: polished writing is not what a UGC script sounds like. A panel without a real-creator judge drifts every script back toward brand copy. The loop: write, let the panel tear it apart, rewrite, run it again. A script that survives is worth spending money on. A script that fails is free to kill. Write 10+ scripts per batch; the panel is cheap next to one generated clip, and discovery needs volume. Record survivors in `ugc/script-bank.md` with format id and judge notes. For synthetic UGC this gate is mechanical: record `script_id` and `judge_verdict` on the manifest asset. The `script_id` must resolve to a recorded row in `ugc/script-bank.md`, and the verdict uses `passed`/`survived` wording — `check:content-assets` blocks a UGC-family generation without both, and rejects a `failed`/`rejected` verdict at any status. Recipe 7 in `tool-recipes/visual-and-motion-production.md` starts at this panel, never at the model.
+
+Daily research habit:
+
+- scout TikTok and adjacent niches for 20-30 minutes
+- save repeated structural patterns, not one-off viral videos
+- look for 3+ unrelated examples with high view counts
+- turn each candidate into 5-8 scripts before judging
+
+Breakout rules:
+
+- 5K views: signal worth inspecting
+- 10K views: stronger signal
+- 50K+ views: likely breakout and bonus trigger
+- one hit: luck
+- 2-3 hits from the same structure, ideally across 2+ creators: format worth scaling
+- after a breakout: clone the hook and structure within 24 hours
+
+Stop rule:
+
+- after 3 months at full cadence, no 100K-ish breakout, no improving format, no install/branded-search lift, and no audience-fit comments means shut down or reposition the channel
+- creator has not posted by day 7: replace
+- creator ghosts, resists feedback, or sends low-effort work in week 1: replace
+- positioning changes every week: freeze it and run 10-15 videos before evaluating
+
+## Tracking And Analytics
+
+Day 0 tracking can stay simple:
+
+- one Google Sheet or equivalent
+- creator
+- account handle
+- post URL
+- date/time
+- format ID
+- hook variant
+- angle
+- CTA
+- product visibility
+- views at 24h, 72h, 7d, 30d
+- comments/theme notes
+- install/branded-search spike notes
+- payout status
+
+This does not replace launch analytics. Use `analytics-attribution.md` for UTMs, creator/referral codes, self-reported attribution, PostHog events, store CTAs, RevenueCat/Stripe conversion, and weekly growth dashboards. Day 0 UGC is intentionally lightweight on attribution, but the launch package still needs measurement once links, waitlists, stores, or purchases are involved.
+
+## Design And Product Implications
+
+Before filming:
+
+- make the app logo or name visible in screenshot-heavy screens
+- identify which 11-star magical moment or complete product experience the video is demonstrating
+- ensure the product moment can be captured in a short screen recording
+- design UI states that are legible inside vertical video
+- avoid tiny copy that disappears in phone-shot footage
+- prepare a truthful demo account/fixture with realistic data
+- mark any generated or mock UI as non-production until the real app exists
+
+Use Higgsfield only for supporting visuals, mascots, backgrounds, hooks, or motion after `DESIGN.md` exists. Do not use generated visuals as a substitute for truthful app UI in store screenshots or product claims.
+
+**Synthetic founder-voice content (before real creators are onboarded).** Marketing Studio supports four UGC-family modes suited to pre-roster founder-led content: `ugc`, `ugc_how_to`, `ugc_unboxing`, and `product_review`. Use these to generate synthetic founder-voice ads during format discovery before a real creator roster is running. Synthetic UGC complements real creator content — it does not replace it; real creators produce authentic human variability, comment-bait tension, and niche credibility that generated video cannot replicate. See the **Soul-Once Founder-Face Ads** recipe in `tool-recipes/visual-and-motion-production.md`. When no founder face should appear, use the **Script-First Believable-Person UGC** recipe (Recipe 7) in the same file. Recipe 7 uses a first-frame image plus a Seedance reference-driven video. Confirm the exact model id through the `higgsfield-generate` skill. Recipe 7 needs no Soul and no avatar.
+
+Guardrails for synthetic UGC:
+
+- Script first. Every synthetic UGC generation uses a script that survived the script-bank judge panel above. The model renders a weak script and a strong script with equal quality. Generation never rescues writing. `check:content-assets` enforces `script_id` and `judge_verdict` on every UGC-family manifest asset.
+- Prompt register. UGC-family modes and Recipe 7 follow the **UGC Realism Prompt Structure** in `tool-recipes/visual-and-motion-production.md`: social context truth, imperfect presence, no resolution. Never use the premium product-ad structure for these modes. Polish makes the viewer identify both the ad and the AI.
+- Believability. Before scaling a synthetic format, one real human who did not make the clip watches it. If they identify it as AI or as an ad, suspect the script first, then the prompt, last the model. Record `passed — <detail>` or `failed — <detail>` in the manifest `believability` field; a done-tier status without an explicit pass fails `check:content-assets`.
+- Automation last. Scale the flow into an automated pipeline only after it passes the believability test repeatedly by hand. Automation of a broken flow makes slop faster.
+- Founder-face content additionally requires a trained founder Soul (`soul-cinematic` variant) and a custom avatar; see `tool-recipes/visual-and-motion-production.md`.
+- Every generation prompt must carry `DESIGN.md` tokens (palette, type mood, shapes, texture, motion energy, banned aesthetics, surface). Generating without this brief is a named failure mode.
+- Gate every generation behind spend confirmation per [`paid-tool-routing.md`](../operations/paid-tool-routing.md); surface current balance first (`mcp__claude_ai_Higgsfield__balance`).
+- Record all generated assets in `CONTENT_ASSETS.md` / `growth/content-assets/manifest.json` with `prompt_brief`, `soul_reference_id`, `avatar_id`, and `virality_score` fields.
+- Synthetic ads must pass brain-activity virality scoring before paid distribution (see **Virality Closed Loop** recipe in `tool-recipes/visual-and-motion-production.md`) — and pass [`quality-lens.md`](../design/quality-lens.md)'s Anti-Generic Checks, which neither the virality score nor the design-token check evaluates.
+- Public posting and paid campaign launch remain founder-gated.
+
+**Long recordings → short clips (`personal_clipper`).** Use `personal_clipper` to cut long founder recordings, podcast appearances, demo walkthroughs, or app explainers into short-form clips for UGC distribution. `personal_clipper` is an MCP-tool capability only (`mcp__claude_ai_Higgsfield__personal_clipper_create` / `_jobs` / `_status`); confirm the exact invocation via the `higgsfield-generate` skill before running. Gate behind spend confirmation per `paid-tool-routing.md`. See the **Master → All Platforms** recipe in `tool-recipes/visual-and-motion-production.md`.
+
+Use Remotion when founder-led or creator-led content needs repeatable local variants: product-demo loops, hook overlays, captions, cutdowns, screenshot motion, before/after frames, or app-preview clips. Load `remotion-content-assets.md` first, record Remotion license status, keep real app UI visible, and add each rendered output to `CONTENT_ASSETS.md` and `growth/content-assets/manifest.json`. Remotion is especially useful before Fastlane because the same composition can render many hook, CTA, locale, and dimension variants without paying for net-new generative media.
+
+## Outputs
+
+Create these when UGC is in scope:
+
+- `growth/UGC_PLAYBOOK.md`: fit decision, 90-day plan, budget, creator profile, sourcing channel, contract checklist, payment model, stop/scale rules
+- `VIRAL_GROWTH.md`: product-led loop, referral/share mechanics, content format lab, monetization timing, analytics proof, abuse controls, and stop/scale rules when creator content is expected to compound
+- `CONTENT_ASSETS.md`: Remotion/Higgsfield/raw-media route, source inputs, license status, render proof, and output registry when rendered or generated media supports the UGC lane
+- `ugc/creator-list.csv`: creators, platform, handle, geography, personal followers, notes, outreach status
+- `ugc/script-bank.md`: format IDs, hooks, scripts, CTA notes, product insertion notes — scripts pass the `no-slop-writing.md` self-check (§6) for claims and patterns while preserving natural creator speech
+- `ugc/tracker.csv` or sheet link: posts, metrics, payouts, format results
+- `ugc/creator-brief.md`: filming rules, disclosure rules, approval flow, account handling, product truth constraints
+- `ugc/weekly-review.md`: current winners, failed hooks, next tests, spend, install/search signals
+- updated `growth/FASTLANE_OPS.md` when Fastlane becomes the generation/scheduling engine after launch
+
+## Post-Breakout Scale Model
+
+Everything above is the Day-0 discovery experiment: 3–5 creators finding a format. Discovery is not the destination — UGC-driven consumer apps at real chart positions run volume, and volume is the mechanism that turns one proven hook into a sustained install stream. Once the breakout bar holds (2–3 hits from the same structure across 2+ creators, with downstream install/paywall/revenue evidence), the roster grows in bands, each a founder-gated budget step:
+
+| Band              | Roster | Weekly video volume | What changes operationally                                                                                                                                               |
+| ----------------- | ------ | ------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
+| Discovery (Day 0) | 3–5    | 9–15                | founder writes scripts, judges hooks personally                                                                                                                          |
+| Proven format     | ~10    | 30–40               | script bank templatizes the winning structure; per-creator briefs; weekly winners review drives the bank                                                                 |
+| Scale             | ~30    | 90–120              | a lead creator or coordinator owns briefing/QA; payment ops move to batch; format variants tested against the control format                                             |
+| Volume            | 75+    | 200+                | agency or marketplace layer runs sourcing/payments; the founder owns format strategy and brand truth only; Fastlane or equivalent handles scheduling/analytics at volume |
+
+Rules that keep scale honest:
+
+- Advance one band at a time; each band's budget is a separate founder gate with the prior band's install-per-video and payback numbers in front of the decision.
+- The breakout bar re-proves per band: a format that worked at 5 creators must hold its install economics at 10 before 30 exists.
+- Fatigue is measured, not feared: when the control format's install-per-video decays across two consecutive weekly reviews, the next variant test is the priority, not more volume on the decaying control.
+- Disclosure, claim review, and brand-truth constraints from the Day-0 contract apply at every band — volume never dilutes the compliance floor.
+
+## Common Failure Modes
+
+- Scaling to 20+ creators before one format works.
+- Asking creators to invent strategy before the founder understands the medium.
+- Optimizing for production polish before hook strength.
+- Changing positioning every few videos.
+- Using affiliate links or rev share too early and forcing unnatural CTAs.
+- Treating one viral video as a format.
+- Forgetting product/logo visibility in the app screens shown on video.
+- Scheduling UGC or creator posts without disclosure, claim review, and founder approval.
+- Spending on paid creator tools or marketplaces without a founder-approved tool decision.