GA4 Data Retention Settings: What You're Losing and How to Keep It

GA4 has a data retention setting that most teams never check. The default is 2 months. That means event-level data older than 60 days is permanently deleted from your property. Not archived. Not compressed. Deleted. And there is no getting it back.

This does not affect your standard reports, which run on aggregated data kept indefinitely. But it kills your ability to build custom Explore analyses on historical data. If you have ever opened an Explore report, set the date range to six months ago, and gotten a blank screen, now you know why.

What the retention setting actually controls

GA4 has two data tiers, and the retention setting only applies to one of them.

Aggregated data (kept indefinitely). The standard reports — Acquisition, Engagement, Monetization, Retention — run on pre-aggregated tables that GA4 maintains forever. Traffic overview, source/medium breakdown, page performance, and conversion counts always show your full history regardless of the retention setting.

Event-level data (subject to retention). Explore reports, funnel analyses, path explorations, segment overlaps, and any custom analysis you build in the Explore workspace run on raw event-level data. This is the data that expires.

The distinction matters because each tier answers different questions. Standard reports tell you what happened at a summary level. Explore tells you why it happened and what specific user segments did. Without event-level data, you lose granular analysis on historical periods.

The two retention options

GA4 standard properties offer two choices: 2 months (the default) and 14 months. There is no 6-month option, no 24-month option, and no "keep everything forever." Google Analytics 360 properties get up to 50 months, but that is an enterprise product.

What you lose at 2 months

With the default retention, several analytical capabilities break in practice.

Year-over-year Explore analysis. You cannot compare this September's user behavior to last September's in Explore because last September's event data was deleted eleven months ago. The standard reports show the year-over-year traffic number, but you cannot drill into segments, funnels, or paths.

Retroactive segment analysis. You discover that mobile visitors from paid social have a terrible conversion rate. You want to investigate whether it has always been this way. With 2-month retention, "history" means the last 60 days. Everything before that is gone.

Funnel and path exploration. If you build a funnel in Explore and set the date range to 90 days, the first month of that range has no data. Cohort analysis from four months ago returns nothing. Understanding how users navigate the site over long periods requires event-level data that 2-month retention deletes.

What you do not lose

Even with 2-month retention, you keep standard report data (all pre-built reports with full history), BigQuery export data (lives in your BigQuery project indefinitely once exported), key event and conversion counts in standard reports, and user properties (user-scoped custom dimensions persist on the user profile regardless of event retention).

The BigQuery safety net

The BigQuery export is the only way to keep GA4 event-level data beyond the retention window. When enabled, GA4 streams every event to a BigQuery dataset. You need a Google Cloud project with billing enabled. The export itself is free from GA4's side. You pay for BigQuery storage and queries, but for most properties the cost is minimal — a property with 100,000 sessions per month generates roughly 3 to 5 GB per month in BigQuery.

What you gain is every event, every parameter, every user property in a queryable SQL database. What you trade is convenience: BigQuery requires SQL knowledge, while Explore is point-and-click. Even if you never query it, the BigQuery export is insurance. The data you did not export is gone when it falls off the retention window.

Keeping your data beyond the retention window

If your reporting workflow relies entirely on standard reports, the retention setting does not affect you much. If you use Explore for funnel analysis, segment comparisons, or cohort retention, the window limits how far back you can look.

The practical steps are: set every property to 14 months on day one during onboarding (alongside timezone setup and enhanced measurement configuration), enable BigQuery export for clients with a Google Cloud presence, and store your own copy of the daily data.

SyncingBeam follows this pattern. The daily pull stores GA4 metrics in its own database at daily grain. When GA4 deletes event-level data after 14 months, the stored daily aggregates remain. Dashboards and reports keep working with full history because they query the stored data, not GA4's retention-limited event store. You lose Explore-level granularity (individual event parameters, user-level paths) once the retention window closes, but you keep the reporting-level aggregates indefinitely. For teams concerned about data freshness and continuity, this daily storage approach is the most practical safeguard.

The one thing you cannot recover

If a property has been running at 2-month retention for a year and you switch it to 14 months today, you do not get the last year's data back. The switch only affects data going forward. Everything older than 2 months at the time of the switch is already deleted.

This is why the setting matters most on day one. The first thing to do when you get access to a new GA4 property — before checking traffic, before reviewing tags — is verify the retention setting is at 14 months. If it is at 2, change it immediately. Every day at the 2-month default is a day of data you will lose in 60 days.