PostHog

# PostHog Review: The All-in-One Product Analytics Suite Built for Engineers **Rating: 4.5/5** --- ## Overview PostHog occupies a genuinely interesting position in the product analytics market: it's an open-source, self-hostable platform that has evolved into a comprehensive product intelligence suite far beyond its original funnel-and-event-tracking roots. Founded in 2020 by James Hawkins and Tim Glaser during Y Combinator, the company has grown rapidly by targeting a specific persona — product-minded engineers and technical founders who are tired of paying enterprise prices for tools that require a data team to operate. What distinguishes PostHog from Mixpanel, Amplitude, or Heap isn't just the open-source angle. It's the deliberate bundling of product analytics, session replay, feature flags, A/B testing, and a data warehouse-native pipeline into a single platform with a single instrumentation layer. The pitch is consolidation: why pay separately for FullStory, LaunchDarkly, Optimizely, and Amplitude when one SDK can power all of them? That pitch is compelling, but it comes with real tradeoffs. Let's dig in. --- ## Key Features **Product Analytics** forms the foundation. PostHog offers event-based tracking with support for funnel analysis, retention cohorts, user paths, lifecycle charts, and session-level dashboards. The query builder is SQL-adjacent enough that engineers feel at home, and the platform supports group analytics — meaning you can analyze behavior at the account or organization level, not just the individual user level, which is crucial for B2B SaaS companies. **Session Replay** is genuinely competitive with dedicated tools. PostHog records user sessions with DOM snapshots rather than video files, keeping storage efficient. You can filter replays directly from analytics events — click a data point in a funnel, then watch the sessions that dropped off. The console log capture and network request capture features make this particularly useful for debugging, not just UX research. **Feature Flags** are first-class citizens, not an afterthought. You can roll out features to percentage-based cohorts, specific user properties, or company attributes. Flags tie directly into experiments, so you can run a proper holdout group without duct-taping together separate tools. **A/B Testing and Experiments** use a Bayesian statistical engine. Experiment results are visualized clearly, with significance thresholds and confidence intervals displayed inline. The setup is straightforward for engineers already using PostHog's SDK. **PostHog Surveys** allow in-app NPS, CSAT, and custom question collection with targeting rules tied to the same user/event data the rest of the platform uses. **Data Pipeline** (formerly PostHog Pipelines) lets you send data to and from external warehouses — Snowflake, BigQuery, Redshift, S3 — using a transformation layer that's becoming increasingly powerful. The newer **Data Warehouse** feature lets you query external sources (Stripe, Salesforce, etc.) directly inside PostHog using a Postgres-compatible SQL interface. PostHog's **SDKs** cover JavaScript, Python, Ruby, Go, PHP, iOS, Android, and React Native, with a well-documented API. The JavaScript snippet is easy to drop in, and autocapture reduces the instrumentation burden significantly for web apps. --- ## Pricing PostHog uses a **usage-based pricing model** with a generous free tier — arguably the most generous in the analytics space for self-serve. - **Free tier:** 1 million events/month for analytics, 5,000 session recordings, 1 million feature flag requests, and 250 survey responses — all free, forever, with no credit card required. - **Paid tiers** kick in as you exceed those limits. Analytics events are billed at roughly **$0.000225 per event** (around $22.50 per 100,000 events) after the free million, though volume discounts reduce this significantly at scale. - Session recordings run approximately **$0.005 per recording** after the free tier. - Feature flag requests cost **$0.0001 each** beyond the free allocation. There's no mandatory "Growth" or "Enterprise" seat-based tier — you pay for what you consume. An early-stage startup genuinely won't pay anything for months. A Series B company processing 50 million events monthly might expect bills in the $300–$800/month range depending on feature usage, which compares favorably to Amplitude's Growth plans, which can reach several thousand dollars for equivalent scale. **Self-hosted (open-source)** is technically free but requires infrastructure management. PostHog recommends their cloud offering for most teams given the operational overhead of running ClickHouse at scale. --- ## Pros & Cons **Pros** - Exceptional free tier — genuinely usable at early-stage without touching a credit card - Consolidation value is real: one SDK replaces four or five tools - Open-source transparency; you can inspect, fork, or contribute to the codebase - Engineer-friendly interface, SQL access, and API-first design - Session replay integrated with analytics is a workflow superpower - Feature flags and experiments are deeply connected to analytics data - Transparent, usage-based pricing that scales predictably **Cons** - UI can feel dense and inconsistent — some features feel more polished than others - Not built for non-technical users; marketers and product managers without SQL comfort may struggle - Self-hosting is complex and resource-intensive (ClickHouse is powerful but demanding) - A/B testing lacks some statistical rigor options that dedicated experimentation platforms (e.g., Statsig, Eppo) offer - Data warehouse and pipeline features are still maturing compared to dedicated ETL tools - Customer support on the free tier is community-only; paid support response times vary --- ## Who Is It Best For? PostHog is purpose-built for **technical teams at startups and growth-stage SaaS companies** where engineers have significant influence over tooling decisions. It's particularly well-suited for: - **Early-stage startups** who want serious analytics without enterprise pricing - **B2B SaaS companies** who need group/account-level analytics alongside user-level data - **Engineering-led product teams** comfortable querying data in SQL - **Companies evaluating open-source** for compliance, data residency, or customization reasons - **Teams tired of tool sprawl** paying separately for analytics, session replay, and feature flags It's a harder sell for **enterprise companies** needing dedicated support SLAs, **non-technical teams** who want a drag-and-drop experience, or organizations running rigorous multi-variate experimentation programs where a dedicated stats engine matters. --- ## Verdict PostHog has executed an audacious product strategy — build a legitimate competitor to multiple category leaders simultaneously — and largely pulled it off. The free tier is genuinely transformative for bootstrapped and early-stage teams, and the consolidation story holds up under scrutiny for most mid-market SaaS use cases. It isn't perfect. The UI reflects the reality of a fast-moving engineering team, and some modules feel less mature than dedicated point solutions. But the overall value equation, especially for technical product teams, is difficult to beat. If you're an engineer who's ever muttered about the cost of your analytics stack, PostHog deserves a serious evaluation. **Best for:** Technical product teams, early-stage SaaS, open-source advocates **Not ideal for:** Non-technical marketers, enterprise experimentation programs **Overall Rating: 4.5/5**