Every platform claiming the same sale — attributed revenue was 312% of actual

The situation

A Canadian pet supplies DTC brand was preparing for a Series A fundraise. Their VC asked for a clean data room including ad attribution data. The CFO pulled numbers from GA4, Meta, and Google Ads — three different revenue figures, none matching Shopify. The discrepancy was $280K/month.

The problem

A full audit revealed: GA4 was double-counting sessions due to duplicate GTM container IDs. Meta pixel was firing on page load AND after a successful form submit — same event, different triggers. Google Ads had three active conversion actions including two legacy ones from a previous agency that never deactivated them. CAPI for Meta was sending events without dedup keys. Total attribution was 312% of Shopify revenue.

What we did

1. Removed duplicate GTM container — identified two container IDs in the site header from two different agency installations
2. Fixed Meta pixel triggers — removed page-load purchase event, kept only dataLayer-driven post-purchase event with CAPI dedup
3. Deactivated 2 legacy Google Ads conversion actions — only "Purchase (Server)" remained as primary
4. Added event_id to all Meta CAPI events using Shopify order ID
5. Implemented Northbeam-style attribution report in Looker Studio showing each channel's revenue with Shopify as source of truth

Why it worked

Attribution inflation is endemic in accounts touched by multiple agencies. Each agency installs their own tracking layer without removing the previous one. The result is compounding overcounting that makes every channel look profitable on paper while actual business performance is opaque. A fundraise-ready data room requires Shopify as ground truth, not ad platform self-reported numbers.