git:20260526.b91d55f to git:20260711.c346837

14 added, 162 removed. Audit A to A.

---

name: ads-attribution

- description: "Cross-platform attribution health audit covering AdAttributionKit (iOS view-through 24h post-impression, WWDC 2025 configurable windows), GA4 attribution models (data-driven vs last-click), Consent Mode V2 enforcement, server-side attribution stitching, MMP integration health, and cross-device / cross-platform attribution. Use when user says attribution audit, attribution model, AdAttributionKit, AAK, view-through attribution, GA4 attribution, Consent Mode V2, conversion window, attribution window, MMP audit, AppsFlyer audit, Adjust audit, Branch audit, Singular audit, cross-device attribution, or cross-platform attribution."

- user-invokable: false

- tested_date: 2026-05-26

- tested_with: claude-code v2.x

+ description: "Audit cross-platform attribution, conversion definitions, reporting windows, GA4, AdServices and AdAttributionKit, MMPs, browser and server events, offline conversions, and platform reconciliation. Use for attribution audit, attribution models, conversion windows, GA4 attribution, MMP review, AppsFlyer, Adjust, Branch, Singular, cross-device attribution, or cross-platform conversion discrepancies."

---

- # Cross-Platform Attribution Health Audit

- Attribution decay is the silent revenue killer of 2026 — Consent Mode V2 EEA

- enforcement (Jul 21, 2025), iOS ATT, SKAdNetwork → AdAttributionKit migration,

- and the death of third-party cookies have moved every advertiser's

- attribution signal toward modeled, server-side, first-party data. Misaligned

- attribution windows or unverified Consent Mode setups will mis-attribute

- 15-40% of conversions and silently waste budget.

- ## Process

- 1. Collect current attribution stack: GA4 property ID, Google Ads conversion

- actions, Meta CAPI config, Apple Ads / AdAttributionKit registration,

- MMP dashboard (AppsFlyer / Adjust / Branch / Singular), any sGTM container

- 2. Read `ads/references/conversion-tracking.md` for the cross-platform

- tracking baseline

- 3. Evaluate attribution health per surface (web, iOS app, Android app,

- server-side)

- 4. Score each surface PASS / WARNING / FAIL

- 5. Generate findings report with cross-channel attribution map and remediation

- plan

- ## What to Analyze

- ### iOS Attribution (AdAttributionKit + ATT)

- - **AdAttributionKit registered** with Apple Ads (post-Apr 10, 2025 cutover);

- registration creates dual attribution with SKAdNetwork (SKAN v1-3)

- - **View-through attribution** active — 24h post-impression view window

- configured on Apple Ads campaigns where applicable

- - **Configurable attribution windows** (WWDC 2025): per-campaign window

- customization audited; overlapping re-engagement windows used for

- subscription / re-acquisition campaigns

- - **Country code in postbacks** (WWDC 2025): enabled if you need geo

- attribution detail

- - **ATT (App Tracking Transparency) opt-in rate** monitored; <30% opt-in

- means heavy reliance on SKAN/AAK + privacy threshold

- - **Privacy threshold awareness** — low-volume campaigns may receive null

- postbacks; campaign consolidation recommended below 1k installs/week

- ### Web Attribution (GA4 + Google Ads + Meta CAPI)

- - **GA4 attribution model**: Data-Driven (default for properties with enough

- data) vs Last-Click — confirm setting is intentional, not residue from a

- pre-2026 migration

- - **Google Ads attribution model**: Data-Driven default; per-conversion

- override allowed but audit any Last-Click overrides for justification

- - **Cross-channel attribution** in GA4: confirm Google Ads, Meta, LinkedIn,

- TikTok, Microsoft are integrated as platforms with consent + auto-tagging

- - **Conversion windows per channel** appropriate to sales cycle:

- - E-commerce: 7-day click, 1-day view

- - B2B / lead gen: 30-90 day click, no view

- - SaaS subscription: 30 day click, 1-day view, plus offline conversion

- import for the activation event

- - **Conversion lag analysis** run quarterly to validate window choice

- ### Consent Mode V2 (EU/EEA + recommended globally)

- - **Consent Mode V2 active** (enforcement began Jul 21, 2025 for EEA/UK)

- - **Advanced Mode** (not Basic) — Basic loses ~25% of EEA conversion signal

- - **700+ ad clicks/day over 7 days per country/domain** threshold met for

- Advanced Mode behavioral modeling to activate

- - **Signal recovery measured**: aim 15-25% conversion recovery vs pre-CMV2

- baseline

- - **Consent banner** correctly reads CMV2 signals (not just GDPR boilerplate)

- ### Server-Side Attribution Stitching

- - **First-party server logs** stored with `user_id` + `event_id` + timestamp

- + platform tag for every conversion event

- - **MMP + first-party stitching**: MMP (AppsFlyer / Adjust / Branch /

- Singular) receives both client-side AND server-side events with shared

- `event_id` for deduplication

- - **Server-side conversion import** to Google Ads (offline conversion import)

- and Meta (CAPI) within 72h of the conversion event

- - **Hash quality**: email / phone fields SHA-256 hashed and lowercased

- before send; cross-platform hashing convention consistent

- - **Deduplication rate** ≥90% (matched event_id between client + server)

- ### MMP Health (Mobile Apps)

- - **MMP integrated** — AppsFlyer / Adjust / Branch / Singular set up before

- any paid campaigns launched

- - **Apple Ads connected as a partner** in MMP dashboard

- - **Post-install events sent back** to Apple Ads, Google UAC, Meta App

- Campaigns, TikTok (enables Maximize Conversions and ROAS bidding)

- - **Event quality**: purchase, subscription_start, trial_start, or other

- revenue events tracked (not just install → registration)

- - **Postback configuration**: SKAN/AAK conversion values map to meaningful

- user actions

- ### Cross-Device & Cross-Platform Attribution

- - **Customer Match** lists synced to Google Ads, Meta (Customer File CA),

- LinkedIn Matched Audiences, TikTok Customer File for cross-device stitching

- - **Enhanced Conversions** active for Google Ads (hashed first-party data,

- ~10% uplift on properly-implemented setups)

- - **CAPI customer_information parameters** (`em`, `ph`, `fn`, `ln`, `ct`,

- `st`, `zp`, `external_id`) sent server-side to Meta

- - **Privacy-safe identifiers** (Customer Match hashes, GAID/IDFA where

- permitted) included in conversion exports

- ### v1.8.0 refresh — iOS 26, Meta March-3 rebuild, Reddit Dual Attribution

- - **iOS 26 ATFP + Expanded Link Tracking Protection** (Sept 15, 2025): Advanced

- Fingerprinting Protection is now default ON in ALL Safari browsing (was Private

- only; ~90% fingerprinting-effectiveness reduction), and Link Tracking Protection

- strips `gclid` / `fbclid` / `msclkid` in ALL Safari. **Server-side conversion APIs

- are mandatory** for any account with >5% iOS Safari traffic. This affects ALL iOS

- attribution pipelines, not just Apple Ads. See `ads/references/compliance-requirements.md`

- (C-iOS-1) and `/ads server-side-tracking`.

- - **Meta March 3, 2026 attribution rebuild**: click-through now counts ONLY link

- clicks; new engage-through column; engaged-view threshold dropped 10s → 5s; default

- windows 7-day click / 1-day engage-through / 1-day view. **YoY data is not comparable

- across the March-3-2026 boundary** — flag this in any reporting view. See `audit-meta`

- checks M54-M59.

- - **Reddit Dual Attribution** (beta May 20, 2026): combines Reddit first-party + MMP +

- SKAdNetwork last-touch in a single Ads Manager view; App Event Optimization GA

- (~22% CPA improvement). Evaluate for accounts running Reddit Max.

- - **AdAttributionKit (no position-level reporting for Apple multi-placement)**: Apple's

- Multiple Search Ad Placements (March 2026) do not expose position-level reporting in

- AdAttributionKit initially — caveat any TAP placement-coverage report.

- ## Key Thresholds

- |--------|------|---------|------|

- | EMQ (Meta Purchase) | ≥8.0 | 6.0-7.9 | <6.0 |

- | Event dedup rate | ≥90% | 70-89% | <70% |

- | Server-side / client-side hit ratio | ≥80% | 50-79% | <50% |

- | ATT opt-in (iOS) | ≥30% | 15-29% | <15% (heavy SKAN reliance) |

- | Offline conversion import latency | <24h | 24-72h | >72h |

- ## Output

- ### Attribution Health Score

- ```

- Attribution Health Score: XX/100 (Grade: X)

+ # Attribution Audit

- iOS (AdAttributionKit + ATT): XX/100 ████████░░ (20%)

- Web (GA4 + Ads + CAPI): XX/100 ██████████ (30%)

- Consent Mode V2: XX/100 █████████░ (15%)

- Server-Side Stitching: XX/100 ████████░░ (20%)

- MMP Health (mobile): XX/100 ███████░░░ (10%)

- Cross-Device / Customer Match: XX/100 ██████░░░░ (5%)

- ```

+ 1. Read the main `ads` contract and normalized account snapshots.

+ 2. Declare the business conversion, value, data window, timezone, currency, and

+ decision the attribution analysis must support.

+ 3. Inventory every browser, server, platform, analytics, MMP, offline, and app

+ attribution source with its identity, counting, deduplication, and privacy rules.

+ 4. Reconcile comparable events and explain differences caused by eligibility,

+ view-through rules, consent, modeled data, conversion lag, thresholds, or scope.

+ 5. Separate measurement quality from platform-reported performance.

+ 6. Return findings, contradictions, confidence, missing evidence, and a measurement

+ improvement plan through the common JSON contract.

- ### Deliverables

+ Do not assume one platform is ground truth, add incompatible reports together, or

+ recommend an attribution model without the operator's decision context.

- - `ATTRIBUTION-AUDIT.md`: Full surface-by-surface findings

- - Cross-channel attribution map (which platform owns which conversion

- windows + which events)

- - Modeled vs reported conversion delta (estimated revenue under- or

- over-attribution)

- - Quick Wins sorted by signal-recovery $ impact

- - Pre-launch checklist for any new platform (Amazon, Apple Ads, TikTok)

- to ensure attribution is wired before spend begins