GA4 standard, BigQuery or your own event layer?
GA4 standard is free and does more than many comparisons claim. This page names the documented limits with their source and says when a custom event layer becomes worthwhile.
In short
Replacing GA4 is rarely the goal. For most websites GA4 standard is enough and the BigQuery export covers the rest. A custom event layer pays off where the event model is part of the product, where raw data is needed beyond the retention window, or where several systems need the same events consistently. This page used to claim that replacing GA4 was the normal case. That was wrong.
There is deliberately no three-year calculation here. GA4 standard is free, GA 360 is quote-only, and an invented list price would be exactly the mistake the previous version made.
What GA4 standard does and does not do
| Criterion | Google Analytics 4 | FW Delta |
|---|---|---|
| Price | GA4 standard is free; GA 360 is quote-only | one-time build plus hosting and BigQuery cost |
| Raw data export | the BigQuery export is available on standard too | raw data sits in your own database from day one |
| Export limit | daily batch export up to 1 million events per day; streaming has no stated volume limit and is billed by volume | your infrastructure is the limit |
| Retention | 2 or 14 months for explorations and funnels; aggregated standard reports are not affected | as long as you choose to keep it |
| Sampling | in explorations above a quota of 10 million events | no sampling, but your own query cost |
| Data control | processed by Google under its terms | processed where you run the systems |
| Attribution | Google models, limited configuration on standard | your model, and your responsibility for it |
| Effort | quick to set up, plenty of expertise available | a project with a measurement plan, build and operations |
Stay with GA4 when
- standard reports and explorations cover the need
- the event volume sits under the export limits
- you want to use Google attribution
- nobody wants to run a data store
- ad platforms are the main consumers anyway
A custom event layer pays off when
- raw data is needed longer than retention allows
- the event model is part of your product
- several systems need the same events consistently
- the daily export limits are hit regularly
- the analysis has to be reproducible and versioned
Documented facts from the Google documentation
GA4 Standard supports the BigQuery export. The daily batch export is limited to 1 million events per day; the streaming export has no stated event volume limit and is billed by BigQuery at roughly USD 0.05 per gigabyte.
As of 2026-08-10 · Source
The GA4 data retention setting offers 2 or 14 months on standard properties. It affects explorations and funnel reports. Standard aggregated reports are not affected by it.
As of 2026-08-10 · Source
Sampling in GA4 explorations starts above a quota of 10 million events for standard properties and up to 1 billion events for 360 properties. It is a query threshold in explorations, not a monthly collection cap.
As of 2026-08-10 · Source
FAQ
Is the BigQuery export really 360 only?
No, and the previous version of this page claimed it was. The BigQuery export is available for standard properties. The daily batch export is limited to one million events per day; the streaming export has no stated volume limit and is billed by BigQuery on data volume.
Do I lose everything after 14 months?
No. The retention setting affects explorations and funnel reports. Aggregated standard reports are not affected. If you need raw data for longer, export it to BigQuery.
What do the 10 million events mean?
That is the quota above which exploration queries on standard properties run on a sample. It is not a monthly cap on data collection.
Do you recommend switching GA4 off?
Usually not. A custom event layer normally complements GA4 rather than replacing it. The ad platforms still need their own signals either way.
Measure first, then decide
The Tracking Check shows whether your problem is GA4 or the measurement in front of it.