GA4 Migration
Most GA4 migrations were done under deadline pressure, by mapping old events to new ones and hoping. The result works well enough to produce numbers and badly enough that nobody trusts them. Rebuilding that setup properly is one of the most common projects we take on.
Two kinds of migration
The first is straightforward: a property that still needs moving, or a new site or app that needs GA4 from scratch. We design the event architecture around your business model rather than around a legacy mapping, and enable the BigQuery export on day one, because the export only collects data from the moment you turn it on and every week without it is history you cannot get back.
The second is more common and more interesting: a migration that happened, technically works, and produces numbers the team argues about. Events named inconsistently across platforms, purchase firing in three different ways on web, iOS and Android, ecommerce parameters half implemented, consent bolted on afterwards, and a dataLayer designed for a site that has since been redesigned twice. Fixing that is not a migration, it is a rebuild with traffic running through it.
How we approach a rebuild
We start with an audit, because you cannot rebuild what you have not measured: what fires, what does not, what disagrees with the backend, and which of the problems are actually costing money versus merely being untidy. That produces a prioritized plan rather than a big-bang rewrite.
Then the design: an event and parameter taxonomy that works the same across web and app, ecommerce implemented to specification rather than approximately, consent wired as a first-class input, and the naming documented so the next person does not invent their own convention. We implement with your developers, QA every release, and validate against BigQuery rather than trusting the realtime report.
Historical continuity gets planned explicitly. Where definitions change, we document when and why, so a year-over-year comparison does not silently compare two different definitions of the same metric. Where old data cannot be reconciled, we say so rather than letting someone discover it in a board deck.
What changes afterwards
Reports that match the backend within a known and explainable margin. One definition of purchase across platforms. Consent behavior you can describe to your legal team. And a BigQuery export you can build product analytics, reporting and AI-ready data on, which is where the real return on this work shows up.
Where this sits
GA4 migration and rebuilds are delivered through Analytics Implementation, almost always starting with a Measurement Audit so the scope is based on evidence rather than assumption.
FAQ
We migrated years ago. Is a rebuild really necessary?
Will we lose historical data?
How long does a rebuild take?
Do you handle app tracking in the same project?