PostHog — product analytics, replay, flags, and errors in one platform

• PostHog

  This page

  PricingPostHog leads

  • Free tier on every product
  • Pay-as-you-go per product
  • No per-seat pricing

• Mixpanel

  PricingPostHog leads

  • Free tier by event volume
  • Growth scales with events
  • Enterprise is custom

• Amplitude

  PricingPostHog leads

  • Free tier by event volume
  • Plus scales with volume
  • Growth & Enterprise custom

1. PostHog — official site
2. PostHog — pricing and free tier
3. PostHog — self-hosting docs
4. PostHog — security and compliance
5. Mixpanel — official site
6. Mixpanel — pricing
7. Amplitude — official site
8. Amplitude — pricing

Free tier

Yes

Pricing summary

PostHog meters each product separately on usage, with a free monthly allowance on every tool and no per-seat charge. The free plan is not a trial: usage stops at the free limits rather than rolling into an invoice, so a card on file cannot produce a surprise bill. Pay-as-you-go removes the ceiling, raises the project and data-retention allowances, and adds email or Slack support, with a billing limit you can set per product. Teams needing SSO, audit logs, and role permissions add a platform package, and serious volume moves to annual plans with volume discounts. PostHog states it deliberately avoids loss-leader pricing and sales calls—confirm current allowances on the official pricing page.

Reviewed on 18 September 2026 · PostHog — pricing and free tier

• Product and web analytics

  Funnels, retention, trends, and paths for product questions, plus a lighter web analytics view for traffic and campaigns—queryable in SQL when the standard reports run out.

• Session replay

  Watch real sessions with console logs and network calls attached, jump from a funnel drop-off straight into the recording, and mask sensitive fields before capture.

• Feature flags and experiments

  Ship behind flags, target cohorts, and run A/B tests whose results are read against the same funnels and retention charts the rest of the team already trusts.

• Error tracking and logs

  Group exceptions with stack traces and search application logs in the same project as the analytics, so a bug and its blast radius share one timeline.

• Surveys and product research

  Target in-app surveys by cohort or behaviour and read the responses beside the usage data they are meant to explain.

• Managed warehouse and SQL

  Query product data next to connectors for external tools, with a SQL editor, saved views, and an API for anything the built-in reports do not cover.

• Open-source core and self-hosting

  The core is MIT licensed and can be self-hosted from a Docker Compose deploy, which matters when a policy demands it—even though PostHog supports the cloud product first.

• Agents, MCP, and integrations

  An MCP server lets coding agents query product data directly, alongside a desktop app, API, webhooks, and a data pipeline for wiring PostHog into existing tooling.

1. 1

   Install with the wizard, not by hand

   PostHog ships a setup wizard that instruments the app for you, so the first event lands before anyone writes tracking code manually.

2. 2

   Answer one question before building dashboards

   Start from a single funnel or retention question and confirm the events are named as expected. Autocapture makes it easy to collect far more than you will ever query.

3. 3

   Set a billing limit per product

   Because each product is metered on its own, cap the ones you switch on so a traffic spike stops at the limit instead of arriving as an invoice.

4. 4

   Move experiments next to the metrics

   Run flags and A/B tests in the same project as the analytics, so a winning variant is judged by the funnel you already validated rather than a separate tool.

5. 5

   Check the self-hosting escape hatch early

   If data residency or an on-prem requirement is non-negotiable, confirm it against the docs before standardising—PostHog states self-hosted instances are unsupported and cover only the free-plan feature set.

• Product managers

  Funnels, retention, and in-app surveys in one place answer “why did the metric move” without waiting on an analyst to build a query.

• Product and full-stack engineers

  One SDK and one project hold analytics, flags, errors, and logs, so instrumenting a feature and debugging it happen in the same tool.

• Growth and marketing teams

  A web analytics view with experiments and session replay covers campaign landing pages without bolting on a separate tag-manager stack.

• Data and platform teams

  SQL access, a managed warehouse, and an MIT-licensed core that can be self-hosted give a path that does not end at a vendor dashboard.

Open source, but cloud is the supported product

The core is MIT licensed and self-hostable via Docker Compose. PostHog states self-hosted deployments are officially unsupported, ship only the free-plan feature set, and receive no tagged releases—every change, including security fixes, lands on master and the latest image.

Usage-based per product, with real free allowances

Each tool is metered separately with a monthly allowance that resets, and billing limits can be set per product. PostHog says the large majority of its customers pay nothing, and the free plan is not a time-limited trial.

No per-seat pricing

Team members are not billed individually; cost tracks events, recordings, flag requests, and similar usage instead, which changes the maths for large read-only audiences.

Data residency and compliance

PostHog Cloud runs in US (Virginia) and EU (Frankfurt) regions, and the company publishes a security page covering SOC 2, GDPR, and CCPA with HIPAA BAAs available for cloud customers. Platform packages add SSO, audit logs, and role-based permissions.

Transparency is part of the sales pitch

The company publishes its handbook, sales manual, and strategy, and ships a public changelog. That is unusual for an analytics vendor and useful when judging how quickly gaps get closed.

Agentic surfaces are new

The platform now markets an inbox that clusters findings from errors, replays, and support into prioritised reports with proposed code changes, plus MCP access for external agents. Verify current behaviour before depending on those surfaces in production.