
Mixpanel
Cloud product analytics, with an optional cookieless mode.

Price and script size are shown as the difference from Amplitude. Prices are compared only between tools that bill in the same unit; where the unit differs, the column shows the unit instead of a difference.
| Tool | Cookies | Where it runs | Data in EU | Free plan | At 100k / month | Script, gzip |
|---|---|---|---|---|---|---|
| Amplitudethe tool you are leaving | Optional | Cloud | Yes | Yes | Freebilled by events | 64.0 KB |
| Mixpanelproduct analytics · checked 17 Sep 2026 | Optional | Cloud | Yes | Yes | Freesame price | 32.7 KB31.3 KB lighter |
| Pendoproduct analytics · checked 17 Sep 2026 | Optional | Cloud | Yes | Yes | On requestbilled by users | 233.1 KB169.0 KB heavier |
| Heapproduct analytics · checked 17 Sep 2026 | Used | Cloud | Yes | Yes | On requestbilled by sessions | 6.01 KBloader |
| PostHogproduct analytics · checked 17 Sep 2026 | Optional | Cloud or self-hosted | Yes | Yes | Freesame price | 95.3 KB31.3 KB heavier |
“At 100k / month” is the monthly price for 100,000 units, as the vendor bills them. Dates under statuses, prices and script sizes show when we checked them. See how we check every value.
This directory holds only 5 product analytics tools, so the list above is short. The 6 below measure something else, mostly pages, visits and traffic sources, but they end up on the same shortlist often enough to be worth the comparison. Their values are checked the same way.
| Tool | Cookies | Where it runs | Data in EU | Free plan | At 100k / month | Script, gzip |
|---|---|---|---|---|---|---|
| Amplitudethe tool you are leaving | Optional | Cloud | Yes | Yes | Freebilled by events | 64.0 KB |
| Google Analytics 4web analytics · checked 17 Sep 2026 | Optional | Cloud | No | Yes | Freesame price | 145.9 KB81.9 KB heavier |
| DataFastweb analytics · checked 17 Sep 2026 | Optional | Cloud | No | No | $19$19 more | 5.13 KB58.9 KB lighter |
| Cabinweb analytics · checked 17 Sep 2026 | None | Cloud | Yes | Yes | $14billed by pageviews | 1.17 KB62.9 KB lighter |
| Koko Analyticsweb analytics · checked 17 Sep 2026 | Optional | Self-hosted | Your server | Partly | Freefree, conditions apply | 0.81 KB63.2 KB lighter |
| Lite Analyticsweb analytics · checked 17 Sep 2026 | None | Cloud | Yes | Yes | €12billed by pageviews | 1.09 KB63.0 KB lighter |
| Matomoweb analytics · checked 17 Sep 2026 | Optional | Cloud or self-hosted | Yes | No | $42billed by hits | 21.6 KB42.4 KB lighter |
Amplitude sells one hosted platform, and the pricing page puts product analytics, web analytics, session replay, experiments, guides and surveys on every plan with unlimited seats. Leaving unbundles several jobs at once: the first table covers the event side, the second covers traffic, and replay, experiments and surveys are separate products again. Check which of those you opened last month, before the columns above decide anything.
The counting model goes with it. Events are the unit on the bill and in the reports, and autocapture in Browser SDK 2 turns page views, sessions, form starts, file downloads and campaign attribution into events by default. A tool billed by pageviews counts something narrower, so the same traffic lands elsewhere on the ladder. Retention differs by plan too: Free keeps one year of data and Plus two, and session replays are kept one month on every plan.
identityStorage set to none nothing is stored, but each visit gets a new device ID, so visits join up only after your code identifies someone.The second table is here because the product analytics category is small, and people leaving one of those tools often land on the traffic side. A web analytics tool reports how many people arrived, where from and which pages they read. Few of them keep user profiles, so cohort retention and per-user journeys are missing or thinner than now.
That move is reasonable when the questions left are about channels, campaigns and pages, and product behaviour is read from your own database. It is the wrong one when experiments, funnels keyed to a signed-in account or replay sit in the weekly routine.
Two tools in the first table read an Amplitude project directly. Mixpanel documents a free migration service for projects under 15 million events, still marked beta: you supply the Amplitude API key and secret, a Mixpanel token and a date range, with a dry run available first. The guide warns that backfilled events count toward the plan in the month Mixpanel receives them. PostHog’s managed migration takes the same pair of Amplitude keys, imports at most a year per job, removes no duplicates and needs a paid plan.
Heap and Pendo document the opposite direction, warehouse export rather than import. Nothing in the second table reads an Amplitude file: Matomo imports aggregated reports from Google Analytics, DataFast reads a Plausible export, and Koko Analytics takes files from WordPress plugins.
Yes. The Export API hands over raw events as JSON, a year per request. Two tools here read the project through the same keys; the rest start counting on install day.
No. Amplitude counts events, including the ones autocapture adds, and recognises devices by an identifier in the browser. Several tools in the second table count pageviews and recognise nobody between visits, and bot filtering differs by vendor. The cookieless page covers the methods.
Charts and dashboards are blocked once the account has passed its event allowance three times, and Amplitude deletes an account that stays over the limit six months after the first block.
The first rows of the tables above, with their measured values.

Cloud product analytics, with an optional cookieless mode.

Cloud product analytics, with an optional cookieless mode.

Cloud product analytics.

Cloud or self-hosted product analytics, with an optional cookieless mode, open source.

Cloud web analytics, with an optional cookieless mode.

Cloud web analytics, with an optional cookieless mode.