Amplitude

Overview

Amplitude is a behavioral analytics platform built for product teams who need to move faster than their instrumentation. Instead of raw event logs, it gives you cohorts, funnels, and retention curves—the three things that actually change product decisions. Who uses it: Product managers trying to justify feature work. Growth teams hunting for the conversion cliff. Analysts who got tired of SQL queries. Anyone shipping to mobile or web where user behavior is noisy and multi-step. The core workflow is straightforward. You instrument your app (SDK or API). Amplitude ingests events. You build a funnel—say, signup → onboarding → first feature use → paid upgrade. The platform shows you where users drop, how many, and which cohorts leak fastest. You can then segment by device, geography, or custom properties, then run a targeted rollout or messaging campaign to the at-risk group. One realistic scenario: You ship a redesigned onboarding flow. Amplitude lets you compare the old cohort's retention curve against the new one in real time. You don't wait for a dashboard refresh or a data team query. You see the delta in hours. That's the operational edge. Pricing is usage-based—tracked events determine your tier. Most mid-market SaaS pay between $2k–$10k/month depending on event volume and feature tier. Startups can start free or cheap; enterprise contracts are custom. The pricing model incentivizes you to be thoughtful about what you track, which is actually healthy. Amplitude competes with Mixpanel (similar event model, slightly different UX), Heap (codeless event capture, less flexibility), and Segment + downstream tools (more modular, more ops overhead). If you're already deep in Snowflake or BigQuery, you might build this yourself. If you need it in weeks, not months, Amplitude wins. Honest limits: The platform is event-centric, so it's strongest for digital products with clear user journeys. If your business is B2B with long sales cycles and few events per user, you'll feel the friction. Retention analysis assumes users return; for one-time transactions, it's less useful. Custom cohort logic can get complex fast—you'll need someone who speaks the query language. And like all analytics platforms, garbage in means garbage out; bad instrumentation makes Amplitude useless. The real value isn't the dashboards. It's the speed of hypothesis testing. Can you answer "Do users who skip the tutorial churn faster?" in 10 minutes instead of 3 days? If yes, Amplitude pays for itself.