Analytics Data Retention: How Long Tools Keep Data

Nobody evaluates an analytics tool on data retention. It sits near the bottom of the feature comparison, somewhere below integrations, and it gets skimmed.

Then eight months in, someone asks whether checkout has got worse since the redesign, or how this Black Friday compares with last, and the answer is that the data no longer exists. There is nothing to do about it at that point. Retention is not a feature you can turn on retroactively.

This is what the major tools keep, sourced to their own documentation and checked on 25 August 2026. Verify anything that matters against the linked pages before you commit, because these numbers change.

What each tool keeps

Google Analytics 4

For a standard property you choose between 2 months and 14 months for event and user-level data. GA4 360 properties can go to 26, 38 or 50 months. Google’s documentation is explicit about what happens next: “When data reaches the end of the retention period, it is deleted automatically on a monthly basis.”

There is a detail in there worth reading twice, because it catches people exactly when it hurts most:

“Large and XL properties are limited to 2 months.” … “When a standard property becomes Large or a 360 property becomes XL, the event-level data retention setting is automatically reduced to 2 months and event-level data older than 2 months becomes inaccessible and is permanently deleted.”

The busier your site gets, the less history you keep. Growth is the thing that takes the data away, and it happens without anybody deciding it should.

The 14-month setting is also not the default, and it is set per property. A great many GA4 properties are quietly running on 2 months because nobody changed it during setup.

Microsoft Clarity

Clarity retains recordings for 30 days from the time of recording. Favourited recordings and a randomly selected sample are kept up to 9 months, per Microsoft’s FAQ.

Thirty days is a genuinely useful window for investigating something you already know about. It cannot answer a question you think of later, which is most of the questions worth asking.

Contentsquare

Contentsquare’s privacy FAQ states that personal data in their solution is kept for 13 months, that session replay data is deleted after 1 month, and that IP addresses are deleted after 3 days. Each of those periods may be extended on request.

The split matters more than either number. Aggregate analysis can look back over a year while the recordings that explain the aggregate are gone after a month. You can see that conversion fell in March. You cannot watch what people were doing when it did.

Air360

Air360 keeps your data. There is no analysis window, no expiry, and retroactivity applies to session replays as well as to aggregate analysis. The comparison you want to run next year is against data being collected now, and it will still be there.

Why the window decides what you can ask

Retention does not limit how much data you have. It limits which questions are answerable, and the questions it removes are a specific and unusually valuable set.

Year-on-year comparison. Seasonal businesses are not really comparable month to month. The question is how this November performs against last November, this sale against the same sale a year ago. Anything under 13 months cannot answer it, and 13 months answers it exactly once before the earlier period expires.

Quarter against the same quarter last year. The standard way any board looks at performance. A 2-month window cannot produce it at all.

Investigating after the fact. This is the one that catches teams. Something looks wrong in the numbers, you trace it back to a change made in the spring, and you want to watch what users were doing on either side of it. With replays expiring after 30 days, the evidence went months before anybody had the question.

Establishing what normal looks like. Deciding whether last week was bad requires knowing what an ordinary week is, across a year of them. Short windows make every fluctuation look like a signal, because there is no baseline to compare against.

Measuring a change you shipped a while ago. Redesigns rarely produce a clean before-and-after within 30 days. Effects show up over a quarter, and by then the before is gone.

The part that is easy to miss

Aggregate data and session recordings usually have different retention periods, and the recordings are almost always shorter.

That combination produces a specific frustration. Your dashboard says conversion on a checkout step dropped 8 percent in March. You want to know why, which means watching people fail at that step in March. The chart goes back thirteen months. The recordings behind it went after one.

When you compare vendors, get both numbers, not the headline one. And check whether the longer number covers raw events or only aggregates, because those are different things and they are often quoted interchangeably.

Why tools expire data at all

It is worth saying plainly that short retention is not vendors being difficult. Storing raw behavioural data is genuinely expensive, and there are good reasons to delete it.

Data minimisation is a GDPR principle: personal data should be kept no longer than necessary for the purpose. Retention limits are a legitimate part of a privacy posture, and a vendor keeping everything forever by default with no configuration would be a different sort of problem.

The distinction that matters is whether the limit is yours to set, in line with a retention policy you decided, or whether it is imposed by a pricing tier or triggered automatically when your traffic grows.

What to ask before you commit

How long are raw events kept, and how long are session recordings kept? Two questions, two numbers. Ask them separately.

Is that a default or a ceiling? A default can be changed. A ceiling cannot, and a ceiling that moves when you cross a traffic threshold is worth knowing about now rather than later.

Does extending it apply retroactively? Usually not. If you switch from 2 months to 14 today, you generally wait 14 months to have 14 months. The right time to fix retention is at setup, which is precisely when nobody is thinking about it.

Does it change with plan or traffic? Growth reducing your history is a real behaviour in at least one major tool, and it is not the one people expect.

Who decides the number, you or the vendor? For a GDPR retention policy, being able to set the period deliberately, and document why, is worth more than any particular value.

The practical advice

If you are already running something with a short window, go and look at the setting today. In GA4 it is one dropdown, it takes a minute, and if it is on 2 months you are two months away from a year-on-year comparison you cannot make.

If you are evaluating tools, ask the two retention questions early rather than at the end. It is one of the few specifications you genuinely cannot fix later, and the cost of getting it wrong is invisible right up until the moment it is not.

---

Sources, checked 25 August 2026: GA4 data retention, Microsoft Clarity FAQ, Contentsquare Privacy FAQ. Vendors change their products and plans, so verify current figures before making a decision on them.