GA4 Data Retention: The 14-Month Limit Explained

GA4 is free, and one of the places you pay for that is retention. A standard property keeps event-level data for two months by default, fourteen at most. If you have ever tried to build a year-over-year exploration and watched it come back empty, this setting is why.

What the retention setting controls

The retention period applies to event-level and user-level data: the raw stream of individual events with their parameters and the user records attached to them. This is the data that powers explorations, funnels built in the exploration workspace, segment overlap, path analysis, and audience evaluation over history.

New properties start at two months. An administrator can raise it to fourteen in the property settings, and it is one of the first things worth checking on any property you inherit, because the default has quietly cost many teams a year of analyzable history. On Analytics 360, the paid tier, considerably longer periods are available.

There is one related subtlety: a setting controls whether the retention clock resets on new activity from the same user. With that enabled, an active user's data keeps rolling forward; a visitor who never returns still expires on schedule.

What it does not affect

Here is the part that confuses people. Open a standard report, set the date range to two years ago, and numbers appear. That is because standard reports run on aggregated tables that GA4 maintains separately, and those are not governed by the event retention window.

So the limit is not "GA4 deletes everything after 14 months". It is narrower and stranger: totals, trends, and the pre-built reports keep working over long ranges, while anything requiring event-level granularity goes dark. You can see how much traffic you had in a month two years ago, but you cannot build an exploration segmenting that month's visitors, run a funnel across it, or ask a question the aggregated tables did not anticipate.

For teams that live in standard reports, the cap is nearly invisible. For anyone doing real analysis, it means every question about the past must have been anticipated and aggregated at the time, which is not how analysis works.

The practical consequences

Year-over-year comparison in explorations is the most common casualty: with a 14-month window you have, at best, a two-month overlap between this year and last. Cohort analysis suffers the same way, since retention curves longer than the window cannot be computed. And any custom funnel or segment you dream up today can only look back as far as the window reaches.

The other consequence is irreversibility. Retention is enforced by deletion, and expired data cannot be recovered by support, by upgrading, or by changing the setting afterwards. Raising the setting from two to fourteen months protects the future, not the past.

How to keep your history

The first move takes a minute: set retention to 14 months now, on every property, so the clock stops working against you.

The durable answer is the BigQuery export. Standard GA4 properties can link to BigQuery at no charge from GA4's side, streaming or batching raw event data into tables you own. Once exported, the data is outside GA4's retention rules entirely and stays as long as you pay BigQuery's storage costs, which are modest at typical site volumes. The catch is skills: analyzing raw event tables means writing SQL, and reconstructing sessions or funnels from the export is genuine engineering work rather than a dashboard exercise.

Some teams instead schedule exports of aggregated report data to spreadsheets or a warehouse. That preserves the numbers they already look at, though like the aggregated reports themselves, it only answers the questions someone thought to ask at the time.

Whichever route you choose, the day to set it up is before you need the history, because none of this works retroactively.

How Analyse handles retention

Analyse keeps your full analytics history on every plan, with no event-level expiry window to configure and no separate warehouse required to see last year properly. Because the stored data contains no personal identifiers by design, as described in the cookieless analytics guide, long retention does not create the privacy liability that pushed GA4 toward aggressive deletion in the first place. Year-over-year traffic questions and long-run retention cohorts work directly in the dashboard against the complete record, without an export pipeline standing between you and your own past.