Broken tracking tends to cluster in the same places. We check each of the following one by one, right down into the implementation. We ourselves have recently handled fixing another site's data mixing into GA4 and introducing server-side measurement in real work.
Double tag firing — is a measurement tag loaded more than once on the same page, inflating page views and events?
Cross-domain data mixing — is traffic from another site or app flowing into the same measurement ID?
Lost conversions — are completed sign-ups and purchases reliably recorded as events?
Double-counted conversions — is a single outcome ballooning to two or three on reload or a back action?
Personal information (PII) leakage — are email addresses or names left in URLs or events?
Tag consolidation / migration mistakes — during a GA4 migration or GTM consolidation, is anything missing or duplicated?
Event design validity — are the actions worth measuring captured without gaps or excess, with consistent naming?
Traffic attribution — are referrers and UTMs applied correctly, and are sources classified as intended?
Internal traffic exclusion — is traffic from your own team, the production company or bots mixing into the metrics?
Enhanced conversions and consent — are enhanced conversions and Consent Mode working as required?
Need for server-side measurement — where browser-side loss is large, is there room to switch to server-side measurement?
