How data retention works
Analytics platforms hold two kinds of data. There are the detailed, user-level and event-level records, and there are the pre-built summary reports assembled from them. Retention settings apply to the detailed records. When the retention period expires, the underlying rows are deleted permanently and only the aggregated reports survive.
In GA4 the setting sits in the property’s admin area, offering a short default and a longer maximum. It is one of the very few settings that quietly destroys something: the deletion is not reversible, and raising the limit later does not restore rows that have already gone. Ad platforms, CRMs and email tools each keep their own separate clocks, so a single customer’s history can survive in one system and vanish from another.
Why data retention matters
The distinction that catches people out is between standard reports and custom analysis. Standard reports keep working long after the retention window, because their totals were calculated in advance. The moment you open an exploration, build a custom segment, or apply a custom dimension to an older period, the tool needs the underlying rows — and if they have been deleted, the report comes back empty or partial.
That is exactly the analysis a business wants when something goes wrong. Comparing this festive season with the last, understanding which cohort of customers stayed, or investigating a channel’s decline all depend on detail that a default retention setting may have already thrown away.
Where data retention goes wrong
The usual failure is silent. Nobody changes the setting when the property is created, nobody notices for a year, and the gap only appears when a comparison is finally attempted. There is no warning and no recovery.
The opposite mistake is keeping everything indefinitely because storage is cheap. Personal data carries obligations under privacy law, and holding identifiable records longer than you have a reason to increases both your legal exposure and the damage a breach would do. Retention should be a deliberate decision with a stated reason, not a default in either direction.
How to act on it
Check the retention setting on the day a property is created, not later, and set it to the longest period your privacy position genuinely supports. Do the same for the ad platforms and the CRM, and write the periods down in one place so nobody has to guess.
Where long-term history matters — seasonal businesses, long sales cycles, anything with repeat purchase — send the raw events to storage you control, such as a BigQuery export, so the platform’s clock stops being your only copy. Then make sure your published policy matches what the systems actually do; a promise to delete data on a schedule that nothing enforces is worse than no promise. Getting these settings right at setup is part of any sound GA4 and tracking setup.