GA4 two years in: what digital teams are still getting wrong

Universal Analytics was switched off in July 2023. Most digital teams are now approaching three years into GA4 and the migration pain has long since settled. But a pattern has persisted: organisations moved to GA4 without fully moving to GA4's model.

The fundamental shift in GA4 is from a session-based model to an event-based one. In Universal Analytics, a session was the unit of measurement. Everything else (pageviews, goals, conversions) was built on top of that. In GA4, every interaction is an event, and sessions are derived from those events rather than the other way round.

This matters because it changes how you should think about measurement. Teams that migrated their UA setup directly into GA4, mapping old goals to new conversions and replicating the same dashboard views, have technically moved platforms but are still asking session-based questions. They are often missing the analytical depth that comes from thinking in events.

The BigQuery export is where this becomes most valuable. GA4's native interface is limited by design. Exporting raw event data to BigQuery lets you ask questions that are not possible in the standard UI: full funnel analysis at the user level, custom attribution models, cross-device journey analysis, cohort behaviour over time. Most teams with GA4 enabled have not turned this on.

The other consistent gap is event tracking quality. GA4 captures a significant amount automatically, but the events that matter for product and commercial decisions (feature engagement, micro-conversions, error states) need to be designed and implemented deliberately. Many GA4 setups have solid session-level data and poor event-level data. That is the wrong way round.

If your GA4 setup is a direct replica of your UA setup, you have moved platforms without changing your measurement capability. The value is in the event model. Start there.