GA4 data retention: change it before your year-over-year report breaks

GA4 keeps the detailed event data behind explorations for only 2 or 14 months on a standard property, and if nobody has opened the setting you may be on 2 months. Standard reports are not affected. Explorations are: any year-over-year view you build there (a segment, a funnel, a path) runs out of data after the retention period. Switching to 14 months takes one minute in Admin > Data collection and modification > Data retention. Do it before the holiday season, so this November is still there to compare with next November.

What GA4 data retention actually controls

The setting decides how long GA4 keeps user-level and event-level data tied to cookies, user IDs and advertising identifiers (Google’s documentation). It does not decide how long your aggregated reports live.

Google is explicit about the split: the retention setting “only affects explorations and funnel reports.” Standard aggregated reports, including comparisons and secondary dimensions, are not affected. If an exploration’s date range goes further back than your retention period, the older part simply shows nothing.

Where you look in GA4	Limited by retention?
Reports (Traffic acquisition, Landing page, Pages and screens…)	No
Comparisons in the date picker, including Previous year	No
Explorations: free form, funnel, path, segment overlap, cohort, user lifetime	Yes
Funnel reports	Yes

So if you only ever read standard reports, your year-over-year numbers survive. The trouble starts the day you need something the standard reports cannot do: last year’s conversion rate for returning visitors only, last year’s checkout funnel, the paths people took from a holiday landing page. Those all live in explorations. For the difference between the two, see what counts as a Google Analytics report.

The options: 2 months, 14 months, or more on 360

Data type	Standard property	Analytics 360 property
User-level data (including key events)	2 or 14 months	2 or 14 months
Event-level data	2 or 14 months	2, 14, 26, 38 or 50 months
Age, gender, interests	Always 2 months	Always 2 months
Google signals data	Follows your setting (2 or 14 months)	Follows your setting, never more than 26 months

All rows come from Google’s data retention page, which also notes that Large and XL properties are limited to 2 months. The same list appears in the Admin API reference, where 26, 38 and 50 months are marked as 360 only and event data only.

Google’s help page lists these options but does not name a default, so do not assume yours is on 14. Open the setting and look.

For a standard property, 14 months is the right answer for almost everyone. There is no cost and no downside for reporting. The only reason to stay on 2 months is a privacy commitment you made to your visitors, in which case keep it and know what you give up.

Why 14 months is just enough for year over year

Fourteen months covers a full year plus two months of slack. That matters for seasonal comparisons. Take an illustrative calendar:

• On 2 months: Black Friday 2026 (27 November) drops out of your explorations around the end of January 2027. Next November, you cannot build a segment or a funnel that compares the two sales.
• On 14 months: Black Friday 2026 stays available in explorations until around early 2028, so the November 2027 comparison works, with a few weeks of margin for a year-end review.

Deletion runs once a month, so the exact cut-off date shifts by a few weeks. Treat both dates as approximate.

How to change GA4 data retention

You need the Editor role on the property (Google’s documentation).

1. In GA4, click Admin.
2. Check you are in the right account and property.
3. Under the property settings, open Data collection and modification > Data retention (Google’s documentation gives this path).
4. Set Event data retention to 14 months. If your screen also shows a separate user data retention option, set it to 14 months too.
5. Decide on Reset user data on new activity (see below).
6. Click Save.

GA4 waits 24 hours before applying the change. During that window you can switch back and nothing is lost.

If you manage more than one property, repeat this for each one. The setting is per property, not per account.

Reset user data on new activity: on or off?

This switch only affects user-level data. When it is on, every new event from a user resets the expiry of that user’s identifier to “now plus the retention period.” When it is off, the data is deleted when the period runs out, whatever the user did since (Google’s documentation).

Turn it on if you want returning customers to keep their history in explorations, which is what most small shops and service sites want for cohort or lifetime analysis. Turn it off if your privacy policy says user data is deleted after a fixed period, because “reset on activity” means a loyal visitor’s identifier never expires while they keep coming back.

What lowering the setting does

Shortening the period is the dangerous direction. When you reduce it, data older than the new period is deleted at the next monthly deletion run. Google’s example is a move from 26 to 14 months: everything older than 14 months goes. There is no undo after the 24-hour window, so double check before you save a shorter value.

Also watch for a change you did not make yourself. Google says a standard property that becomes a Large property, or a 360 property that becomes XL, is limited to 2 months, with older event data removed and email warnings sent to administrators. If a high-traffic site suddenly loses exploration history, check its property tier.

To see whether someone on the team changed retention, open Admin > Property change history, as described in the traffic drop checklist.

Keeping more than 14 months: BigQuery export

If you need raw event data beyond 14 months on a standard property, the workaround is to export it to BigQuery, where you decide how long it stays. Some limits to know before you rely on it:

• It starts when you link it. Google’s setup guide says data starts flowing within 24 hours of linking and describes no way to export the months before that. Link it now if you think you will want it.
• Standard properties have a daily export limit of 1 million events (Google’s documentation). Most small sites stay far below it.
• The free sandbox deletes tables after 60 days. BigQuery’s sandbox expires all tables and partitions after 60 days and caps you at 10 GiB of storage over the project’s lifetime, a quota that deleting data does not give back (BigQuery docs). For long-term storage you need to enable billing and then update the expiration times on the dataset and on tables and partitions already created, as Google’s sandbox page advises, otherwise your archive quietly deletes itself.
• Someone has to write SQL. The export is raw events, not reports. For a solo marketer this is only worth it if you already have, or can borrow, someone comfortable with BigQuery.

For most small sites: set 14 months, rely on standard reports for long-range year over year, and skip BigQuery until you have a question that needs it.

A five-minute retention check

1. Open Admin > Data collection and modification > Data retention and set event data to 14 months.
2. Decide on the reset switch and save.
3. Write down last year’s key numbers for your peak weeks from a standard report with Previous year comparison (Google’s help on comparisons), so you have them whatever happens to explorations. The Black Friday analytics checklist lists which ones.
4. List the explorations you will want to repeat next year (a checkout funnel, a returning-visitor segment), so you know which ones depend on the 14 months.
5. If you want raw history past 14 months, link BigQuery today and fix the sandbox expiration.

A record of every week that does not depend on retention

Retention is about the raw data. The other half of a year-over-year story is knowing what happened at the time: which week dipped, what you thought caused it, what you changed. That part gets lost even with perfect retention, because nobody writes it down.

This is where our product, Dialytics, helps. It posts a weekly digest of your GA4 numbers in Slack every Monday, with a short explanation of what moved and the one point worth your attention, and each digest stays in your Slack channel and in the digest history on your Dialytics dashboard (Slack’s free plan only shows 90 days of messages and deletes anything older than a year, per Slack’s help). It also checks every morning for a day that is unusual for that weekday and alerts you only then. It does not replace the retention setting: change that first.