Amplitude

I tested Amplitude on a SaaS analytics workspace with 90 days of historical event data covering signup, activation, and feature adoption funnels for ~12k monthly active users. The cohort I had in mind was product managers and growth teams running quantitative product analytics, distinct from Mixpanel as a peer competitor and from Plausible (lightweight web analytics, very different surface).

The funnel and retention reporting is where Amplitude clearly earns its category-leader status. Building a 5-step activation funnel with conversion windows, segmenting by acquisition channel and plan tier, and overlaying a 12-week retention curve took about 8 minutes once the events were instrumented properly. The Pathfinder feature for tracing actual user journeys through the product surfaced a feature-discovery dead end I had not seen in any dashboard summary: 31 percent of new users hit settings within the first session and bounced from there, which became its own deep-dive investigation. Amplitude AI for natural-language queries actually returned reasonable cohort definitions on simple questions like "users who activated within 7 days but did not return in week 2."

What surprised me was how punishing the schema discipline has to be. If your event taxonomy was loose at instrumentation time (mixed casing, properties with the same semantic meaning under different names), the analysis surface multiplies that mess and the team ends up debating whether sign_up vs Sign Up vs SignUp are the same event in dashboards. The chart-load latency on workspaces with 30+ million monthly events also gets noticeable: I had several queries spinning for 30-60 seconds on broader date ranges, which kills the exploratory mode that makes a tool like this valuable. Pricing tier jumps are aggressive once you cross the free 10M-events ceiling, and several mid-tier features (data governance, identity resolution) gate behind enterprise plans.

Amplitude is the right pick for product teams that already have clean event instrumentation and need deep funnel, retention, and pathing analysis at scale. Skip it if your team is at the "just starting product analytics" stage (Mixpanel is more forgiving on schema), if your needs are 90 percent web traffic dashboards (Plausible or GA4 are radically simpler), or if your event volume cannot justify the price-tier jumps that hit hard above 10M monthly events.