Walled Garden Reconciliation

Every platform claims credit. The reconciliation problem, sources of disagreement, tactical and strategic resolution, and the data clean room future.

Why walled gardens are different

Meta, Google, Amazon, TikTok, Apple, Microsoft, LinkedIn — these platforms operate as "walled gardens": they collect user data within their ecosystems, run their own ad auctions, run their own attribution, and don't share raw user-level data with advertisers. The advertiser sees aggregate reports.

This is privileged data. The walled gardens see things no one else sees: cross-device journeys within their identity graph, view-through conversions on their inventory, content-engagement signals tied to ads. That privileged view makes their reported attribution different from any external measurement.

The reconciliation challenge: walled-garden numbers don't reconcile with each other or with your unified analytics. Every platform claims more credit than reality allows. Cross-platform measurement is hard precisely because of this structural reality.

The reconciliation problem

A simple example: a user sees Meta ad on Tuesday, Google search ad on Thursday, then buys on Friday. Meta's attribution: 100% credit (last 7-day view + click). Google's attribution: 100% credit (last click). Both claim the conversion. Add it up: 200% of revenue attributed.

Multiply by every major platform. Add MTA in GA4 that splits credit differently. Add incrementality tests that show neither channel may have caused the purchase. The numbers don't reconcile because they're measuring different things from different vantage points.

Sources of disagreement

• Different attribution windows. Meta 7-day click + 1-day view; Google 30-day click default; TikTok 7-day click + 1-day view by default but configurable.
• View-through vs click-through. Some platforms count view-through aggressively; others click-only.
• Last-click within platform. Each platform takes last-click credit for users it touched, regardless of other channels.
• Cross-device unification differences. Each platform's identity graph; different match rates.
• Modeled conversions. Platforms model conversions they can't observe. Methodology and accuracy varies.
• Self-serving bias. Each platform's attribution methodology is designed to make them look effective.
• Data freshness. Late-arriving conversions; report refresh lags differ by platform.
• Geographic and consent differences. EEA conversion under-reporting; iOS ATT signal loss varies.

Tactical reconciliation

• Standardize attribution windows. Where you can configure, set consistent lookback windows across platforms.
• Deduplicate with CAPI and dedup keys. Server-side conversion APIs with unique transaction IDs avoid platform double-counting.
• Use UTM tagging discipline. Consistent campaign tagging across platforms; GA4 sees the same campaign across.
• Compare against unified analytics. GA4 (or warehouse equivalent) is one source of truth for cross-channel; reconcile platforms against it, not against each other.
• Adjust for view-through. Subtract view-through-only conversions from Meta/Google reported numbers when comparing to click-attribution sources.
• Recognize that 100% reconciliation isn't the goal. Sources of truth differ by purpose; reconciliation should be directionally tight, not exactly matching.

Strategic reconciliation

• MMM as the umbrella. Aggregate-level MMM doesn't care about platform-reported attribution; it estimates channel contribution from sales data + spend data + controls.
• Incrementality tests for ground truth. Independent of platform attribution; quarterly tests on largest platforms.
• Triangulation reporting. Don't expect platforms to agree; report their numbers alongside MMM and incrementality.
• Discount factor for inflated platform numbers. Some mature programs apply explicit discount factors (e.g., divide platform-reported ROAS by 1.5x to approximate iROAS) based on calibration tests.

Data clean rooms

The infrastructure being built to enable cross-walled-garden measurement without breaking privacy:

• Amazon Marketing Cloud (AMC). Amazon's clean room; combine Amazon ads, DSP, and your first-party data for cross-platform analysis.
• Google Ads Data Hub (ADH). Google's clean room; query Google ad data joined with your warehouse data.
• Snowflake Data Clean Rooms. Cross-organization data sharing without exposing raw records.
• LiveRamp Habu, Datavant, Disney's clean room, etc. Industry-specific clean rooms.
• Use cases: Cross-platform audience overlap, cross-platform incremental lift, full-funnel attribution with first-party data joins.

Advanced playbook

• Build a master reconciliation dashboard. Pull platform-reported numbers, GA4 numbers, MMM estimates, incrementality results into one view. Surface where they agree and disagree.
• Calibration multiplier per platform. Run incrementality tests; calculate platform-reported ROAS / iROAS ratio. Use that ratio as ongoing discount factor.
• Cross-platform conversion paths. Use GA4 path analysis or warehouse SQL to surface multi-platform journeys. Quantify cross-platform contribution patterns.
• Clean room investments. AMC, ADH, Snowflake DCR as platform learnings infrastructure. Slow ramp; long ROI.
• Vendor measurement layer. Tools like Northbeam, Triple Whale, Rockerbox unify platform data with reconciliation logic built in.
• Stakeholder education. Train channel teams on why platform numbers overstate. Without education, optimization decisions optimize against inflated numbers.
• Quarterly reconciliation review. Where are gaps growing? What changed? Methodology audit.
• Walled-garden-specific incrementality. Test Meta within Meta's ecosystem; test Google within Google's. Then test cross-platform via geo holdouts.
• UTM and dedup discipline at scale. Automated validation in CI for campaign tagging; transaction-ID dedup keys for CAPI.
• Communications discipline. Don't share platform-reported ROAS as ground truth in board materials. Use triangulated numbers.

Common mistakes

• Adding up platform-reported ROAS as if conversion credit is additive; gross over-counting.
• Trusting one platform's number over another based on which is highest.
• No CAPI implementation; platform attribution depends on browser-pixel-only data.
• No standard UTM tagging across platforms; cross-platform attribution impossible.
• Expecting platforms to match GA4; they won't and shouldn't exactly.
• No incrementality testing; platform-reported numbers treated as truth.
• Clean rooms ignored; missing the cross-walled-garden analytical infrastructure.
• No discount factor applied to platform-reported ROAS in strategic planning.
• Stakeholders not educated; channel optimization rewards inflated metrics.
• No documentation of reconciliation methodology; reporting drifts.

Operating checklist

• Reconciliation dashboard comparing platform-reported, GA4, MMM, incrementality
• CAPI implemented across all major platforms with dedup keys
• UTM tagging standards documented and enforced
• Standardized attribution windows where configurable
• Quarterly incrementality test rotation calibrating platform numbers
• Platform-specific discount factors documented and applied
• Clean room access (AMC, ADH, or industry-specific) at least piloted
• Stakeholder education on walled-garden over-counting
• Communications policy: triangulated numbers, not platform-reported, in strategic materials
• Quarterly reconciliation review with methodology audit

---

Part of the Attribution & Measurement series.