PostHog
Web AnalyticsOpen-source product analytics platform
Replaces Google Analytics, Mixpanel
3 self-hosted Amplitude alternatives
Amplitude is a commercial product analytics platform that helps teams understand user behavior through events, funnels, and cohorts. It is aimed at product and growth teams. These open-source apps let you replace Amplitude with software you host and control yourself.
Best self-hosted Amplitude alternatives
Last reviewed Oct 11, 2026 · 607 words
If you want Amplitude's funnels, retention and cohorts on your own servers, PostHog is the answer, and it is not a close call. It covers product analytics, session replay, feature flags and experiments in one MIT-licensed codebase. Be honest about the cost first: the directory rates it Hard to set up with a 4 GB RAM floor, and the catalogue notes that the self-hosted edition is scaled down from PostHog's own cloud. If you only use Amplitude to count page views and signups, you are about to over-engineer, and a lighter web analytics tool will serve you better.
What you gain and what you give up
Gained: event data that never leaves your infrastructure, no per-event bill, and tools Amplitude sells separately (session replay and feature flags) running against the same event stream. You can also query the underlying data directly instead of through a vendor export.
Given up: managed scale and polish. Amplitude's governance features, its data taxonomy tooling and the comfort of someone else being paged when ingestion falls over all stay behind. A self-hosted PostHog is a stateful analytics system with several moving parts; the catalogue's con, "heavy to self-host", is an understatement on small hardware. Teams with modest traffic usually cope fine; teams sending very large event volumes should price the hardware before they price the licence saving.
Why not the other two options listed?
The other two cards on this page are infrastructure, not Amplitude replacements. Rakam offers SQL over collected events, but its repository shows no pushes since 2021 in the catalogue, which rules it out for anything new. The behavioural data pipeline listed second (Scala, Apache-2.0, Snowplow's collector) gathers well-structured event data into a warehouse, but it has no funnel or retention UI of its own; you would still need a BI tool such as Metabase or Superset on top. That is a valid architecture for a data team, and a long detour for a product team.
| PostHog | Event pipeline | Rakam | |
|---|---|---|---|
| Directory's setup estimate | Hard, 4 GB | Hard, 2 GB | Hard, 2 GB |
| Funnels and retention UI | Yes | No, needs a BI layer | Reports via SQL |
| Maintained | Very active | Active repo, older releases | Inactive since 2021 |
How the migration actually goes
The tracking calls are the easy part. Amplitude's track('Signed Up', props) maps almost one to one onto PostHog's capture('Signed Up', props), and identify calls map the same way, so keep your existing event names and property schema and the dashboards you rebuild will mean the same thing. Ship both SDKs side by side for a few weeks, then compare daily counts for your top ten events before removing Amplitude.
History is the hard part. Amplitude has an export API that returns your raw events, and PostHog accepts events with their original timestamps through its ingestion API, so a backfill is possible. It is a scripting job, though, and charts, cohorts and saved funnels do not migrate at all. Most teams I would advise to backfill only the last year, rebuild the five dashboards people actually open, and keep an Amplitude export archived for anything older.
What I'd do
Run PostHog on a dedicated machine with more than the 4 GB minimum, dual-track for a month, then cut over. If your honest use of Amplitude is traffic and conversion counts rather than behavioural cohorts, skip PostHog and read the PostHog vs Umami comparison and ditching Google Analytics, which cover the lighter tools.