We audit GA4 setups as part of every new client onboarding. In the last 18 months, 68% of them have had critical tracking failures — broken conversion events, missing UTM attribution, inflated sessions from bot traffic, or key user actions that were never instrumented in the first place.
The universal analytics sunset forced migration, but it did not force most teams to actually understand what they had migrated into. GA4 is a fundamentally different product — event-based, not session-based — and treating it like the old UA is how you end up with dashboards that look fine but measure nothing meaningful.
The five mistakes we see constantly
1. Using the auto-migration and calling it done
Google's auto-migration tool moved your basic structure across. It did not migrate your custom goals, your funnel configurations, your cross-domain tracking, or your custom dimensions. If you hit "migrate" and moved on, you are almost certainly missing critical data.
2. Not instrumenting micro-conversions
In UA, you probably tracked a handful of goal completions. GA4's event model lets you track everything — and you should. Scroll depth, video completions, file downloads, form starts (not just completions), internal search queries. These micro-conversions are the leading indicators that tell you where the funnel is leaking before it shows up in your revenue numbers.
3. Ignoring the session_start / first_visit events
GA4 does not count sessions the same way UA did. The definition changed. If you are comparing GA4 session numbers to historical UA numbers without understanding the definitional shift, you will draw completely wrong conclusions about growth or decline.
4. Not setting up enhanced measurement correctly
GA4's enhanced measurement sounds like a free win. It partially is — but it also tracks outbound clicks to your own subdomains as exits if cross-domain tracking is not configured. We see this break attribution for multi-domain setups constantly.
5. No data freshness monitoring
GA4 data can stop flowing without any obvious error in the interface. We have seen clients make budget decisions based on three weeks of missing conversion data because no one noticed. Build a simple monitoring check — a daily email alert or a Looker Studio dashboard with a data freshness indicator.
What a clean setup looks like
Properly configured GA4 with GTM, clean event taxonomy, a documented custom dimension schema, cross-domain tracking tested and verified, conversion events mapped to actual business outcomes, and a Looker Studio dashboard built for humans not data engineers. That is the target. It takes 2–4 weeks of focused work to get there. The analytical clarity it delivers is worth far more than that investment.