Start from your metering unit: events (Amplitude, Mixpanel, PostHog, Statsig), sessions (LogRocket, Fullstory, Heap, Hotjar), monthly active users (Pendo, Userpilot) or page hits (Matomo). Then decide whether you need only analytics, or also replay, guides and experiments, because bundles price those separately. PostHog, Matomo and Countly can be self-hosted, which is a different support and cost decision from their managed clouds.
Free includes 1M events per month. Growth starts at $0 and scales with volume (with yearly billing, 1.5M events is $100 per month). Enterprise is custom.
Usage-based. The page example of 1.5M events at $100 per month assumes yearly billing ($1,200 per year, saving $480 versus monthly). Enterprise is by quote.
When to choose it
Self-serve event analytics on a volume-based Growth plan.
What to confirm
Growth is volume-based, so spend moves with event count. The page example of 1.5M events at $100 per month on yearly billing is about $140 monthly.
First 1M events per month free (also 5K recordings and 1M flag requests). Usage-based after that with no minimum spend or annual contract.
This is PostHog Cloud. Self-hosting is a free MIT-licensed Docker Compose hobby deploy that PostHog says is officially unsupported, with no paid support (posthog.com/docs/self-host).
When to choose it
Engineering-led teams that want product analytics, session replay, feature flags and experiments in one tool with usage-based pricing and no annual contract.
What to confirm
Managed PostHog Cloud is usage-based per product, so each product (events, recordings, flags) is metered separately. The self-hosted MIT-licensed Docker Compose hobby deploy is officially unsupported and has no paid support.
Documented capabilities
Product analytics with the first 1M events free per month
Session replay with 5,000 free recordings per month
Feature flags with 1M free requests per month
Usage-based pricing with no minimum spend or annual contract
Events, sessions, MAU and page hits grow at different rates as a product scales, and overage behavior (soft cap, auto-upgrade, hard stop) changes the bill.
Pull last month's event, session and MAU counts and run them through each vendor's calculator, then read what happens when you pass the plan limit.
Free-tier ceiling
Free plans differ widely in events, sessions, projects and retention, and some drop key features like replay or experiments.
Compare the free-tier limits and retention window with your real volume before assuming you can stay free.
Autocapture versus tracked events
Autocapture speeds adoption but can inflate event counts and cost, while hand-tracked events take engineering time but give cleaner data.
Run a week of autocapture and compare the event volume with a planned tracking plan.
Privacy, consent and data residency
Session replay and analytics process user data, so masking controls, consent handling and regional hosting matter for compliance.
Check the default masking, consent options and available data-hosting regions in the vendor's docs and security pages.
Self-hosted versus managed analytics
Self-hosting gives data control but moves scaling, upgrades and support to your team, and some vendors state the self-hosted build is unsupported or feature-limited.
Read the self-hosting docs for support, scale limits and license terms, and compare with the managed plan at your volume.