Migration path · from Google Analytics 4
Move from GA4 to Amplitude without losing your history
GA4's sampling, retention limits, and interface drove you here. The migration itself is an event-schema problem - and that is a solvable, testable engineering task.
Why do teams leave Google Analytics 4?
- Sampling on the reports you present to leadership
- Data retention windows that quietly delete your history
- An interface your PMs refuse to self-serve in
- Thresholding and cardinality limits hiding your own data
How do Google Analytics 4 concepts map to Amplitude?
| GA4 concept | Amplitude equivalent |
|---|---|
| Event + parameters | Event + event properties |
| User properties | User properties |
| Sessions (session_start) | Sessions (configurable definition) |
| Conversions / key events | Any event in a funnel or metric |
| Audiences | Behavioral cohorts |
| Explorations | Charts (funnel, retention, segmentation) |
| BigQuery export | Warehouse destinations / exports |
Concept mapping - names differ, and definitions differ in places (sessions especially). The migration tracking plan pins each pair down explicitly before any data moves.
What actually migrates - and what shouldn't?
Your event schema migrates; your bad habits shouldn't. GA4 setups accumulate auto-tracked noise and parameter sprawl, so we start by extracting the events your team actually queries, map those onto a clean Amplitude taxonomy, and leave the debris behind. The result is usually a smaller, sharper event set than the one you left.
Historical data is a strategy decision, not a copy-paste: options range from a BigQuery-side archive you query for look-backs, to backfilling key aggregates, to a clean break with a documented baseline. We scope which fits your reporting obligations.
How do you know the new numbers are right?
Parallel run. Both systems collect for an agreed window while we reconcile the core metrics and document every legitimate definitional difference (GA4 sessions are not Amplitude sessions). Cutover happens when the deltas are explained, not when the calendar says so.
Related: the implementation sprint that follows the mapping, and the head-to-head comparison if you are still deciding.
Frequently asked questions
Can I import historical GA4 data into Amplitude?+
Partially - and only history that exists in event-level form. GA4's BigQuery export accumulates from the day it is linked, with no backfill of earlier data, so what you can move is bounded by when someone turned the export on. If it's been running, Amplitude can bulk-ingest the transformed events; if not, your pre-export history exists only as GA4 aggregates. Often a BigQuery archive for look-backs plus a clean Amplitude start is the cheaper, more defensible call. We scope this in the first week.
How long does a GA4 to Amplitude migration take?+
Typically 4-8 weeks end to end: one to two weeks for schema extraction and the migration tracking plan, one to three for instrumentation, and a two-to-four-week parallel run before cutover. A single product surface lands at the short end.
Should we keep GA4 running after moving to Amplitude?+
Usually yes, in a reduced role - GA4 remains useful for acquisition and ads-ecosystem reporting, while Amplitude owns product analytics. We define which questions live where so the two never compete for the same metric.
Get a free, scored audit of your Amplitude instance
Send us read-only access and get a scored findings report within 48 hours: taxonomy health, duplicate events, governance gaps, and the three fixes with the highest data-trust payoff. No commitment.
Request the free audit