GA4 BigQuery Backfill: Can You Export Historical Data?

No. The GA4 BigQuery export starts on the day you link the property and only sends data from then on. As of October 2026 there is still no option in GA4 to send past events to BigQuery, and Google's help is explicit about the related cases: data that has been exported cannot be re-exported, days missed while the export is paused for going over the limit are not reprocessed, and data from an export interrupted by an invalid payment method cannot be sent again.

So if you linked GA4 to BigQuery in March, your event-level history in BigQuery starts in March. Everything before that exists only inside GA4, in aggregated form, and only for as long as GA4 keeps it.

The only "late" data the export does handle is events that arrive a little after they happen: Google keeps updating each daily table with events that come in up to three days late. That is not a backfill, just a delay.

There is one official route that does reach into the past, and it is often mistaken for a backfill of the export. The BigQuery Data Transfer Service has a Google Analytics 4 connector. According to the Google Cloud documentation in October 2026:

• It loads reporting data from the Google Analytics Data API, not events. You get tables such as p_ga4_Events, p_ga4_TrafficAcquisition, p_ga4_LandingPage and p_ga4_PagesAndScreens, with dimensions and metrics already aggregated.
• It runs daily, with a refresh window of 4 days by default and up to 30.
• You can run a backfill for earlier dates, as far back as the data retention policy of the source allows.
• Running the transfer has no cost; normal BigQuery storage and query prices apply.
• Its limits: totals for distinct users and sessions may not be accurate or match GA4, custom properties are not supported, and a custom report allows one per transfer, with up to 9 dimensions and 10 metrics and no custom dimensions or metrics.
• It does not work in the BigQuery sandbox, which does not support the Data Transfer Service.

On retention, GA4's own help says its data retention setting affects explorations and funnel reports, not standard aggregated reports. Even so, run a test backfill on your property before promising anyone a start date.

In other words, the connector gives you historical reports in BigQuery. It does not give you historical events: no event_params, no user_pseudo_id and no session IDs to join with your CRM.

If you need a continuous history, the practical answer is usually a combination: aggregated tables from the connector or the Data API for the period before the link, and the event export from the link onwards, with a clear date where one ends and the other starts.

Two things to keep in mind:

• The metrics will not line up perfectly at the join. Users and sessions in GA4 reports are estimates, and the connector documentation itself warns those totals may differ. Use a period where you have both sources to measure the difference, and write it down.
• Keep the long comparisons at the aggregated level. A year-on-year chart of sessions or key events by channel works across the join. A user-level funnel or a match with CRM leads only works from the day the event export started.

For the event side, see the GA4 export schema and how to count sessions. For comparing channels over time, keep in mind that the connector and the export do not use the same attribution: how attribution models differ.

Very little at company-website volumes. BigQuery storage in the EU and US multi-regions costs $0.02 per GiB a month for active data and $0.01 for data not modified in 90 days, with the first 10 GiB each month free, according to the Google Cloud pricing page in October 2026. Google estimates about 600,000 GA4 events per GB, so a site with a few hundred thousand events a month adds well under a gigabyte.

The real cost is queries and the work of turning events into something the business can read. That is what our BigQuery consulting covers, and it is why the daily export limit is worth checking before it becomes a gap.