How to build multi-vendor receipt joining + offline attribution end-to-end

The 4-skill bundle on the attribution-ops agent

One agent. Four coordinated skills. The Join + Resolve + Attribute + Audit bundle runs above the POS + offline-conversion + identity-graph + CDP + causal-inference substrate and writes one canonical per-receipt + per-attribution record.

Compliance overlay

Five anchors run per-receipt + per-attribution before any commit. The first anchor is operationally distinctive: post- iOS-14.5 attribution collapse + per-platform offline-conversion ingestion + per-PII hashing intersect per-vendor identity-match rate attestation.

FAQ

What is multi-vendor receipt joining + offline attribution end-to-end — and what is the post-iOS-14.5-attribution-collapse-times-PII-hashing problem distinctive to this skill?

A DTC ecommerce + multi-location operator processes per-receipt transactions across Stripe + Shopify + WooCommerce + Magento + Square + Toast + Clover + Lightspeed + Aloha POS. Each receipt needs to be joined back to the upstream paid acquisition signal (Meta + TikTok + Snapchat + Pinterest + LinkedIn + YouTube + Google Ads + Microsoft + Apple Search Ads) and the upstream organic / direct / email / SMS signal. The four-skill bundle on the attribution-ops agent — Join, Resolve, Attribute, Audit — sits above the POS + offline-conversion + identity-graph + CDP + causal-inference substrate and writes a per-receipt + per-attribution canonical record. The operationally distinctive anchor: Apple App Tracking Transparency iOS 14.5 IDFA + Google Privacy Sandbox have collapsed deterministic attribution. Per-platform offline-conversion ingestion (Meta Offline Events + Google Ads Offline Conversions + TikTok Offline Events + Pinterest CAPI + LinkedIn Offline Conversions + Snapchat CAPI + Microsoft Ads Offline) requires per-PII hashing (SHA-256 + bcrypt + Argon2 + scrypt + Identity Match per-platform) before upload. Per-platform identity-match rate attestation. Plus replication-crisis discipline (Brier + ECE + Rosenbaum + Bonferroni + BH FDR + Holm + Storey + pre-registration + ASA p-value statement) + GDPR Article 22 + ECOA disparate-impact when attribution drives credit/insurance/employment decision + EU AI Act Article 22 + Annex III + SOX 302/404/906 + PCI DSS v4.0 Requirement 3.4.1 + 12.10 when payment-card data.

Why do Stripe + Shopify + Square + LiveRamp + Salesforce Data Cloud + Funnel.io + AppsFlyer + Adjust break at multi-vendor receipt-joining-with-offline-attribution scale post-iOS-14.5?

Each commerce vendor ships per-tenant flat receipt + transaction primitives. Each identity vendor ships flat identity graph. Each offline-conversion API ships flat upload. Each MMP ships flat attribution. None coordinates per-receipt cross-vendor identity-resolution with per-PII hashing (SHA-256 + bcrypt + Argon2 + scrypt + Identity Match per-platform) + per-vendor identity-match rate attestation. None implements causal-inference rigor (PyMC + Stan + NumPyro + bambi Bayesian + CausalML + DoubleML + EconML + Synth + CausalImpact + GeoLift + propensity-score + RDD) with replication-crisis discipline (Brier + ECE + Rosenbaum + Bonferroni + BH FDR + Holm + Storey + pre-registration). None gates per-attribution against Apple ATT iOS 14.5 + Google Privacy Sandbox + per-platform CAPI + per-platform offline-conversion attestation + GDPR Article 22 + ECOA Reg B disparate-impact + EU AI Act Article 22 + SOX + PCI DSS v4.0. None writes a per-receipt + per-attribution audit trail with regulatory-defense retention. The four-skill bundle Join + Resolve + Attribute + Audit sits above the commerce + offline-conversion + identity + CDP + causal-inference substrate — it does not replace it.

How does Join + Resolve work across multi-vendor receipt + post-iOS-14.5 identity-resolution?

Join runs per-portfolio per-banner per-location per-receipt extraction from Stripe + Shopify + WooCommerce + Magento + BigCommerce + Square + Toast + Clover + Lightspeed + Aloha + NCR POS + Stripe Terminal + Square Terminal + Shopify POS. Per-receipt: transaction ID + timestamp + amount + per-line-item + per-payment-method + per-tender + customer PII (email + phone + name + address) when available + per-loyalty-ID + per-anonymous-session-ID. Per-receipt PII hashing (SHA-256 + bcrypt + Argon2 + scrypt + per-platform Identity Match algorithm) before any upload to per-platform offline-conversion ingestion (Meta Offline Events + Google Ads Offline Conversions + TikTok Offline Events + Pinterest Conversions API + LinkedIn Offline Conversions + Snapchat Conversions API + Microsoft Ads Offline Conversions). Resolve runs per-portfolio cross-vendor identity-resolution via LiveRamp + Salesforce Data Cloud + Adobe Real-Time CDP + Twilio Segment + Treasure Data + mParticle + Tealium + RudderStack + Segment Personas + Lytics + Iterable Persistence Layer + Acxiom + Experian + Epsilon Abacus + Truthset identity graph. Per-vendor identity-match rate attestation (Meta hash-matching + Google Customer Match + TikTok Identity Resolution + LinkedIn Matched Audiences + Snapchat Audience Match). Per-receipt cross-channel attribution graph: upstream paid acquisition + organic + direct + email + SMS + push + in-app + display + voice + offline.

What does Attribute + Audit do?

Attribute runs per-portfolio per-receipt attribution through causal-inference rigor: PyMC + Stan + NumPyro + bambi + brms Bayesian + CausalML + DoubleML + EconML + Synth + CausalImpact + GeoLift + Differences + Matching + propensity-score + regression-discontinuity-design. Per-receipt attribution method selection: time-decay + position-based + linear + first-touch + last-touch + data-driven + custom. Per-attribution replication-crisis discipline: per-test power-analysis + per-test minimum-detectable-effect + per-test pre-registration of analysis plan per Center for Open Science + AsPredicted + protocols.io + OSF + per-test-family multiple-comparisons correction (Bonferroni + BH FDR + Benjamini-Yekutieli + Holm-Bonferroni + Hochberg + Sidak + Storey q-value) + Rosenbaum bounds + E-value + ASA statement on p-values 2016/2019 + Brier + ECE + reliability + isotonic + Platt + Beta calibration. Per-platform Conversions API match-rate attestation. Gate runs 5 anchors per-receipt + per-attribution before any commit. (1) Apple ATT iOS 14.5 + Google Privacy Sandbox + per-platform CAPI + per-platform offline-conversion + per-PII hashing + per-vendor identity-match rate attestation. (2) GDPR Article 6/7/22/28/30 + EU-US DPF + SCCs + UK IDTA + CCPA + CPRA + 18-state + LGPD + DPDP + PIPEDA + Quebec Law 25 + COPPA + Washington MHMDA + Texas SCOPE + state biometric + Illinois BIPA + Texas CUBI + Washington biometric. (3) ECOA Reg B 12 CFR 1002 disparate-impact when attribution drives credit/insurance/employment + 4/5ths rule + Title VII + ADA + Fair Housing + EEOC AI Guidance 2024 + NYC Local Law 144 AEDT + FCRA Section 604/615 + GLBA Safeguards Rule + CFPB Circular 2022-03. (4) Replication-crisis discipline + EU AI Act Article 22 + Article 26 + Article 50 + Article 13/14/15 + Annex III + Article 6/27 FRIA + DSA Article 26 + Article 28 + Article 30 + DMA. (5) SOX 302/404/906 when public-company + COSO + Exchange Act 13(b)(2) + FASB ASC 606 revenue recognition + ASC 350 + IRC Section 263 + SEC Reg S-K + SEC Form 10-K + FTC Endorsement Guides + Fake Review Rule + Section 5 + Pfizer 1972 + MARS + CFPB UDAAP + state UDTPA + Lanham + PCI DSS v4.0 Requirement 3.4.1 + 12.10 when payment-card data. Audit writes a per-receipt + per-attribution WORM canonical record: per-receipt POS source + per-PII hash + per-vendor identity-match attestation + per-platform CAPI attestation + per-platform offline-conversion upload + causal-inference method + calibration metrics + per-test pre-registration + multiple-comparisons applied + Rosenbaum + E-value + per-anchor gate-pass + AI-ML attribution decision + ECOA retrospective + FRIA. Storage: AWS S3 Object Lock + Azure Blob immutable + GCS + Wasabi WORM. Retention: 7-year FTC + 7-year IRS Section 263 + 7-year SOX + 6-year SEC + 3-year FINRA + 4-year ECOA + 7-year PCI DSS + GDPR Article 30 + EU AI Act Article 12 + SOC 2 CC7/CC8.

What does this skill connect to on the attribution-ops agent and across the swarm?

On the attribution-ops agent: marketing-mix-modeling + attribution-analysis + customer-journey-tracking + per-location cross-channel attribution rollup for franchise reporting (sibling build-pillar) + root-cause attribution sketch for multi-location KPI diagnosis (sibling build-pillar). Across the swarm: per-location creative fatigue detection (#581 same post-iOS-14.5 + per-platform CAPI substrate) + buyer-state-aware BANT scoring (#568 same ECOA + EU AI Act Article 22 + replication-crisis substrate) + false-positive suppression (#574 same replication-crisis substrate) + multi-stream severity routing (#578 same SEC 8-K + PCI DSS substrate) + versioned customer history for multi-location DSAR (#584 same GDPR + identity-resolution substrate). Build-pillar siblings: tiered pre-filter deterministic gates for AI content compliance + marketing AI autonomy profile configuration + per-vertical compliance overlay. Commercial-pillar parent: /attribution-analysis.

What does the 6-workstream pre-engagement-baseline reporting cycle look like for this skill?

Every two weeks during the Tier 3 Fractional CMO with AI Swarm engagement, six workstreams report against the pre-engagement baseline. Workstream 1: per-portfolio POS + offline-conversion coverage — POS vendors enrolled + per-platform CAPI + Apple ATT + Google PS attestation + per-platform offline-conversion ingestion. Workstream 2: Join receipt-extraction flow — per-receipt volume + per-PII hashing coverage + per-platform Identity Match upload + per-vendor identity-match rate. Workstream 3: Resolve cross-vendor identity-resolution — LiveRamp + Salesforce Data Cloud + Adobe RT-CDP + Treasure Data + identity-graph attestation + per-vendor identity-match rate. Workstream 4: Attribute causal-inference flow — Bayesian + CausalML + DoubleML + EconML + Synth + CausalImpact + GeoLift + RDD applied + per-test pre-registration + Bonferroni + BH FDR + Brier + ECE + Rosenbaum + E-value. Workstream 5: Regulatory-defense audit coverage — Apple ATT + per-platform CAPI + offline-conversion + per-PII hashing + GDPR Article 22 + ECOA + EU AI Act Article 22 + replication-crisis + SOX + PCI DSS v4.0. Workstream 6: FBC feedback-loop pattern-learning — per-receipt realized-vs-predicted attribution + per-platform match-rate reconciliation + per-vertical pattern shifts.

Related reading

• attribution analysis
• Sibling build-pillar: per-location creative fatigue detection (#581 same post-iOS-14.5 substrate)
• Sibling build-pillar: false-positive suppression (#574 same replication-crisis substrate)
• Sibling build-pillar: versioned customer history for DSAR (#584 same GDPR + identity-resolution substrate)
• Fractional CMO with AI Swarm
• AI Readiness Assessment