How long does each analytics tool keep your data? The aggregate-retention vs raw-data trade-off
If you're searching for how long your analytics tool keeps your data, the short answer depends on a distinction most tools never explain: raw per-visit data (one row per individual visit, often tied to an identifier) versus aggregate stats (daily totals — views, visitors, top pages). GA4 keeps user- and event-level data for 2 months by default, 14 months at most on a standard property; Plausible keeps aggregate stats 3 years (Starter/Growth) to 5 years (Business); Fathom keeps data forever; Simple Analytics keeps 30 days on its free plan and full history on paid; and Simplytics deletes raw visit rows nightly but puts no scheduled expiry on daily totals while an account holds them.
Here's the insight that ties it together: the raw data is the clock. Per-visit rows are the part with a retention limit, because they're large and they're a privacy liability. Aggregates are smaller, although text dimensions such as page paths and event names may still identify someone. Simplytics inverts the usual arrangement — it throws the raw rows away every night and keeps held daily totals without a scheduled expiry. That makes long-term history practical while an account holds it, without keeping every raw visit.
Raw data vs aggregates — why the difference decides everything
A raw visit record is one row per visit: a timestamp, a page, often a visitor identifier, plus whatever dimensions were captured. These rows are how a tool can later answer questions like "show me every visit from Germany on mobile in March." They're also the rows that make a dataset personal data — they can describe an individual's path through your site.
An aggregate is a count: "2026-03-14 — 412 views, 318 visitors, top page /pricing." No per-visit rows, but some dimensions can still contain identifying text. A year of aggregates for a busy site is kilobytes. A year of raw rows is millions of records.
So retention policies almost always set a clock on the raw rows, not the aggregates, for two reasons: storage cost, and liability. The longer you hold raw per-visit data, the more you store and the more personal data you're responsible for. That clock is why GA4's user-level history runs out at 14 months even though its aggregated standard reports persist — once the raw events age out, you can no longer build a new Exploration over that period.
What each tool keeps
| Tool | Raw visit data | Aggregate / report history | Notes |
|---|---|---|---|
| Simplytics | Deleted nightly (worst case under 48 h — see the arithmetic below) | No scheduled expiry while held | No visitor cookies; $12/year |
| GA4 | User/event-level: 2 months default, 14 months max (50 months only on Analytics 360) | Aggregated standard reports persist; Explorations bounded by raw window | Free; cookies + consent |
| Plausible | None (privacy-first, no raw personal data) | 3 years (Starter/Growth), 5 years (Business) | From $9/mo |
| Fathom | Forever | Forever | From $15/mo; cookie-less |
| Simple Analytics | — | 30 days (free), full history (paid) | Paid from $20/mo; cookie-less |
| Matomo | Configurable — can keep raw logs; auto-deletion is opt-in (self-hosted default keeps them) | Processed reports can be kept indefinitely | Full control; Cloud ~€29/mo for 50k hits |
A few of these deserve their honest credit. Fathom also offers indefinite aggregate history — a similar long-term benefit, subject to its own account terms. Plausible's 3–5 year window is genuinely generous, and it also aims to minimize raw visitor data. Matomo gives you the most control of any tool here — you decide what raw data to keep and for how long. The point of the table isn't that one tool retains more than everyone else. It's that the tools differ on what they retain and what that costs you, in money and in liability.
How Simplytics' nightly deletion actually works
Simplytics records raw per-visit rows (events and sessions) during the day for one job: deduplicating unique visitors within that day. Once a local day is complete, those rows are wiped. The in-progress local day is kept so that cross-timezone dedup doesn't break, and the job asks each site's own timezone what day it is there before deleting anything. Work it through. The job runs at 00:00 UTC. For a site east of UTC by O hours, its local day has already flipped by then, so the day just finished is deleted that same run and the first row of that day is 24 + O hours old — 24 h at UTC, 29½ h in India, 38 h at the far eastern edge. For a site west of UTC the day genuinely has not ended at 00:00 UTC, so it waits one more run: 48 − |O| hours, which is 40 h in California and at most 47 h an hour west of Greenwich. Worst case, anywhere on earth: under 48 hours.
This page said “under ~48 hours” from June and it was not true until 2026-09-15. The figure had been copied from a code comment written for UTC, while the job compared site-local dates against a UTC one — so it could not delete a day until that day was over in every zone, which made it 48 + O instead: 53½ h in India, 62 h at UTC+14. The job now wipes against each site's own local date, and the sentence is true for the first time. After that, what remains is the aggregate daily totals: views, visitors, sessions, and every dimension, kept without a scheduled expiry while the account holds them.
One nuance worth stating plainly, because funnels need some notion of a path: the page-journey records used for funnels (the "transitions" table) drop the visitor identifier by the second nightly run — within two nights, and the records themselves are deleted after 90 days. (That one is a rolling 24-hour rule rather than a day boundary, so a visit at 23:00 UTC loses its identifier 25 hours later and one at 00:30 UTC takes 47½. We wrote “once their day is over” until 2026-09-15; it is the second night, not the first.) So even the funnel data carries no per-visitor identity beyond a couple of days. Data is stored in the EU (Warsaw).
The result is a tool that can show you a five-year traffic trend while retaining far less visit-level history. The trend lives in aggregates; the raw rows are deleted within 48 hours. A page path or event name inside an aggregate can still identify someone if a site sends it.
The honest trade-off: what you give up
This is the cost, and it's real. Because Simplytics deletes raw rows, you cannot slice the past in ways you didn't plan for. If you didn't aggregate a dimension at collection time, that detail is gone — you can't retroactively add a new segment to historical data, run an ad-hoc query over individual past visits, or build a brand-new custom report on old raw events.
Tools that keep raw data buy you exactly that flexibility. GA4 (within its 2–14 month window) and Matomo (for as long as you choose to retain logs) let you go back and ask a question of the past that you didn't think to ask at the time. If "re-analyze last quarter's raw visits along a dimension I just thought of" is a workflow you need, a pre-aggregated tool is the wrong choice and you should know that before you switch.
What you get in exchange is long-term trend history with almost no privacy surface area: you keep the thing most site owners actually look at — visitors, pages, referrers, channels over time — without keeping the rows that make a breach or a subject-access request painful.
Where the price comes in
Retention is partly a billing decision. Longer raw-data history costs the vendor storage, and that shows up in tiers. Plausible's 5-year window sits on its $19/mo Business plan; GA4's longer windows mean its data model and, ultimately, Analytics 360; Simple Analytics' full history starts at $20/mo; Fathom's forever-aggregates start at $15/mo; Matomo Cloud is around €29/mo for 50k hits.
At $12/year, Simplytics puts no scheduled expiry on daily totals while your account holds them, with 50,000 pageviews/month, up to 12 websites, a ~1.8 KB gzipped script (how that compares), and a 30-day free trial, no card. The honest framing: Simplytics is among the cheapest paid, hosted privacy-friendly analytics tools without a scheduled expiry on held daily totals, at $1/month billed yearly — a lot of functionality for a lot less money than the $9–15/month of the privacy-first competitors, achieved partly by keeping the least raw data, not the most.
The short version
Most tools put a clock on raw per-visit data because that's the part that's large and personal. GA4's clock is 2–14 months; Plausible holds aggregates 3–5 years; Fathom keeps everything forever; Simple Analytics' free tier holds only 30 days; Matomo lets you decide. Simplytics deletes raw rows nightly and keeps held daily totals without a scheduled expiry — long-term trends with less raw data — for $1/month billed yearly.
If long-term trend history matters more to you than re-querying old raw visits, that trade is a good one. If it doesn't, keep a tool that retains raw data — just know what window it gives you. The comparisons lay out the full picture, including where each tool wins: Simplytics vs Google Analytics, vs Plausible, vs Fathom, and vs Matomo.
Competitor retention figures and pricing reflect each vendor's published docs as of June 21, 2026. If a vendor changes its policy, email us and we'll update this post.