PostHogProduct analytics, session recordings, feature flags, and A/B tests in one platform, with an open-source core.
Repo
PostHog
Product analytics, session recordings, feature flags, and A/B tests in one platform, with an open-source core.
The most complete open-source analytics stack: funnels, session recordings, feature flags, experiments, and surveys share one system, so product-led GTM teams stop stitching five vendors together. The catch is that the self-hosted free build is deliberately limited, and serious volume pushes you to their paid cloud.
Strength
Watch real session recordings to see exactly where signups stall, then test fixes with built-in A/B testing, all against one shared data set.
Strength
Generous free tier, and an AI assistant that answers product questions in plain language and builds the dashboards for you.
Watch out
Running it on your own servers is officially a hobby option with vague scaling past about 100K events per month. Real volume means PostHog Cloud and its event-based pricing.
LanguagePython
LicenseMIT (ee/ directory proprietary)
Self-hostYes (deprioritized, small-scale)
Managed cloudYes (PostHog Cloud)
First release2020
Stars38.1k
What it does
- Funnels, retention, trends, and cohorts that show where visitors convert and where they abandon
- Session replay recordings, so you watch real users get stuck instead of guessing from charts
- Feature flags that switch features on for selected segments and roll them back instantly
- A/B testing with clear statistical readouts, so page and onboarding changes get measured properly
- Surveys, error tracking, and imports from tools like Postgres and BigQuery
- An AI assistant that answers questions in plain language, builds dashboards, and helps debug broken tracking
What it can’t do
- Match its own cloud product when self-hosted; the free build is not meant for high-volume production use
- Claim fully open licensing; some enterprise features are proprietary and stripped from the public mirror
- Run lightly; self-hosting requires serious backend machinery and a real IT commitment
- Do marketing attribution or ad-spend ROI end to end; it tracks product behavior, not campaign performance
- Load invisibly; its tracking code can slow page load when installed early, especially with recording on
Pros & cons
Pros
- Analytics, replay, flags, and experiments share one data set and one bill
- Open-source core with a huge active community and constant releases
- Strong free tier, with public pricing, roadmap, and company handbook
- The AI assistant genuinely speeds up getting answers out of your data
- Open data formats give you an exit route if pricing or the company changes
Cons
- Self-hosting is deprioritized by design and does not scale past modest volume
- DIY deployment demands ongoing IT commitment and real backend muscle
- Some enterprise features sit behind a proprietary license
- Cloud pricing climbs quickly at high event volume
- The breadth takes more learning than a single-purpose analytics tool
Recently fixed
Pulled from the repo’s closed issues and PRs. A live project fixes things in public; a dead one doesn’t.
- feature feat(data-warehouse): pick a materialized view as a destination or workflow trigger
- feature feat(flags): measure replica staleness on empty override reads
- fix fix(experiments): Reject compound aggregate expressions in HogQL metrics
- feature feat(alerts): render the evaluation history chart with quill charts
- fix fix(metrics): aggregate across series instead of raw samples
- fix fix(data-imports): resolve relative next-page URLs in REST source paginators
- feature feat(data-warehouse): sync id-like binary columns as hex strings
- fix fix(postgres): decode inet/cidr columns to strings
What the community says
I'm normally not a fan of AI being shoved into products but I love it in PostHog. I can ask it a product question, ask it to slice and dice data, and it does a really good job of sifting through my events, doing analysis, building dashboards. It's normally boring tedious work to do that stuff in PostHog so I'm impressed.
I recently set up Posthog in production on 3 apps and Posthog's MCP and integrated AI chat were very helpful for onboarding: verifying integrations, setting up and recommending dashboards, debugging an event not being tracked properly.
The other thing it does is provide confidence against vendor lock-in. I know that if I buy PostHog, and they go out of business or increase their prices, I have an out.
Watch it work
Real tutorials and reviews from YouTube, not ours.
9 sourcesWhere this page’s facts came from. Open the receipts.
- 1GitHub — PostHog/posthog repo
- 2GitHub API — repo metadata
- 3GitHub — recently closed PRs/issues
- 4PostHog LICENSE (MIT Expat + ee/)
- 5PostHog homepage
- 6Hacker News — PostHog discussion (leros, dmix, paulddraper)
- 7Hacker News — self-host/FOSS critique (xyst)
- 8Hacker News — open-source framing critique (JimDabell)
- 9YouTube — What is PostHog? (official)