How to Build an Amplitude Retention Chart You Can Trust
Retention curves lie more often than they're wrong. This covers the return event, the start event, interval choice, and the two findings people keep misreading.
Amplitude is product analytics for event-level user behaviour: funnels, retention curves, cohorts and path analysis. The Starter plan is free up to 50,000 monthly events; Growth pricing starts around $61/month and scales with volume.
Amplitude is a San Francisco company that listed on Nasdaq in 2021. It ingests events from an SDK, builds the usual models (funnel, retention, segmentation, user paths) without SQL, and now layers AI features on top โ session replay, anomaly detection and natural-language questions to the data.
Where Google Analytics answers “how many”, Amplitude answers “which behaviour predicts someone staying”. That distinction decides whether you need it at all.
The free tier is genuinely usable. 50,000 events/month covers a small app at maybe 1,000 to 2,000 daily active users, depending on how chatty your tracking is. If your team fires 30 events per session, that budget disappears in a week.
Pricing is not transparent. Growth starts around $61/month for 300,000 events, but beyond that you talk to sales. Teams regularly report renewal quotes that jump 2x to 4x after the first year, which is worth negotiating hard up front.
Data hygiene decides everything. Amplitude does not fix bad event naming, and there is no built-in tracking plan enforcement on the free plan. Half the disappointing Amplitude deployments I have seen were really tracking-schema disasters.
Mixpanel is comparable and often cheaper at similar volume. PostHog bundles analytics, replay and flags in one open-source stack with self-hosting. Pick Amplitude if the team is non-technical and wants polished charts plus AI answers; pick PostHog if you want data ownership and one bill.