Run Meta, Google, TikTok, and Snapchat simultaneously and your platforms will report roughly double your actual revenue. Not because anyone is lying — because every platform credits itself for the touches it can see, with windows and identity models designed to flatter. Cross-platform attribution is the discipline of reconciling these claims into numbers a budget decision can survive. Here is the methodology we maintain for multi-platform clients.

Step one: enforce UTMs like accounting policy

UTM hygiene is the cheapest analytics upgrade in marketing: lowercase everything, one delimiter convention (underscore vs hyphen — pick one forever), a governed naming sheet, and automated linting that blocks non-conforming links from launch. When UTMs are consistent, GA4 (or your warehouse) becomes the neutral scorekeeper platforms cannot spin.

Step two: capture click IDs at the source

  • Meta: fbp/fbc cookies, captured with the lead and stored in the CRM.
  • Google: gclid, captured on form submit and passed through to CRM — never relying on session storage alone.
  • TikTok: ttp/ttclid; Snapchat: sc_click_id.
  • These identifiers power offline conversion matching (see CAPI patterns) — the technical backbone of platform-truth reconciliation.

Step three: calibration factors per platform

Compute weekly: platform-reported conversions ÷ source-of-truth conversions, per platform, per objective. A healthy stack shows stable ratios (Meta claiming 1.4x truth, Google 1.2x, etc.). Direction and stability matter more than the absolute gap — a ratio that drifts signals tracking breakage or dedupe failure before the dashboards do. Publish the factors beside every platform-reported number in internal reporting.

Step four: triangulate, don't collapse

  1. Source of truth (CRM/orders): what the business actually banked.
  2. GA4 with disciplined UTMs: the neutral journey view.
  3. Platform data: optimization signal only — directional, self-interested, indispensable.
  4. Quarterly incrementality tests: the only causal layer — see advanced measurement.

The weekly reconciliation workflow

One hour, every week: export source-of-truth revenue/leads by UTM; pull platform-reported conversions; compute calibration factors and flag drift >15%; investigate anomalies before Monday's budget meeting. Teams that run this stop having attribution arguments — they have discrepancies with known explanations and known owners.

Should we buy an attribution platform?

Solve taxonomy and click-ID capture first — most teams get 80% of the value free. Attribution platforms add value when journey complexity exceeds GA4's patience and budget justifies a dedicated tool.

What discrepancy is normal?

Platform totals at 1.5-2x source-of-truth are common and reconcilable via calibration. What should alarm you is change — a factor that moved 30% week over week is a tracking bug, not a market shift.

Reconciled numbers or recalibrated instincts — pick one. Get the system.