Definitions written by implementers
The product analytics glossary
Sixteen terms your team should mean the same thing by. Each entry: a liftable definition, the method, one worked example, and where it connects to practice.
Cohort Analysis
Cohort analysis groups users by a shared characteristic - most often the week or month they first used a product - and tracks each group's behavior over time. Instead of one blended average, you see whether users acquired in March retain better than users acquired in January, which isolates the effect of product changes from the effect of who happened to sign up.
Funnel Analysis
Funnel analysis measures how many users progress through each step of a defined sequence - signup flow, checkout, onboarding - and where they abandon it. Its output is a conversion rate per step, which turns 'checkout feels broken' into 'we lose 38% of users at the payment-method screen.'
Product Analytics Platform
A product analytics platform is software that collects user interaction events from web and mobile products and turns them into behavioral analyses: funnels, retention curves, cohorts, and experiment readouts. It answers what users do inside a product - as distinct from web/marketing analytics, which answers how users arrive at it.
Session Replay
Session replay reconstructs an individual user's session as a watchable recreation of their clicks, scrolls, inputs, and navigation. Where event analytics tells you 38% of users abandoned a step, replay shows you individual users hesitating, mis-clicking, or fighting a broken control on that step - the qualitative why behind the quantitative what.
North Star Metric
A north star metric is the single metric a product team treats as its primary measure of delivered customer value - chosen so that moving it requires genuinely helping users, not gaming a number. Classic examples: weekly active teams for a collaboration tool, orders delivered on time for a marketplace, hours listened for audio.
Retention Analysis
Retention analysis measures whether users continue returning to a product after they start, typically plotted as the percentage of a cohort still active N days or weeks later. It is the single most honest health metric a product has: acquisition can be bought, engagement can spike, but retention only moves when the product delivers repeatable value.
Tracking Plan
A tracking plan is a versioned document - usually a spreadsheet or a data-management entry - that defines every event a product tracks: name, trigger, properties with types and allowed values, owner, and the question it exists to answer. It is the contract between the people who ask product questions and the engineers who instrument the answers.
Event Tracking
Event tracking is the practice of recording discrete user actions - signup completed, report exported, checkout finished - as structured data: an event name, a timestamp, a user identity, and properties describing context. It is the raw material of all product analytics; every funnel, cohort, and retention curve is a query over tracked events.
Mobile App Analytics Tools
Mobile app analytics tools track how users behave inside iOS and Android apps - screens viewed, actions taken, retention over time - alongside mobile-specific concerns like app version adoption and crash correlation. The category spans product analytics platforms with mobile SDKs (Amplitude, Mixpanel), attribution tools (AppsFlyer, Adjust), and store/performance consoles.
Product Analytics Metrics
Product analytics metrics are the standard measures of product health derived from event data: activation rate (new users reaching first value), retention (users still active after N periods), engagement (frequency and depth of core actions), conversion (completion of key flows), and the north-star tree that ties them to strategy.
Amplitude Analytics
Amplitude is an event-based product analytics platform - 'an AI analytics platform' in its own current framing: products send it a stream of user events, and teams analyze that stream through funnels, retention curves, behavioral cohorts, and experiments - without writing SQL. It also ships session replay, heatmaps, AI agents, and an MCP integration that lets AI assistants query the data directly.
Behavioral Analytics Tools
Behavioral analytics tools analyze sequences of user actions over time - not just what happened, but in what order, how often, and by whom. The category includes event-based product analytics platforms (Amplitude, Mixpanel, PostHog) and adjacent qualitative tools (replay, heatmaps); its defining capability is following an identified user across sessions and platforms.
Product Analytics
Product analytics is the practice of collecting and analyzing behavioral event data to understand how people actually use a product - and to change it accordingly. Its core loop: instrument events, establish baselines for activation, retention, and conversion, form hypotheses about the gaps, ship changes, and measure again.
Event-Based Analytics
Event-based analytics models user activity as a stream of discrete, timestamped events - each with a user identity and descriptive properties - rather than as pageviews grouped into sessions. Any behavior becomes analyzable: a funnel is a sequence query over events, retention is a recurrence query, a cohort is a filter on event history.
Event Taxonomy
An event taxonomy is the naming and structural system for a product's analytics events: the convention every event name follows (commonly Object-Action, like 'Report Exported'), the standard property vocabulary shared across events, and the categorization that keeps hundreds of events findable. It is the grammar that makes event data readable by humans, and increasingly by AI assistants.
Monthly Tracked Users (MTU)
Monthly tracked users (MTU) counts every unique user an analytics platform records in a calendar month - including anonymous visitors who never sign in. It matters chiefly as a billing unit: analytics vendors have priced by MTUs, and the count is always higher than your MAU, because 'tracked' includes everyone the SDK saw, not just engaged users.
Get a free, scored audit of your Amplitude instance
Send us read-only access and get a scored findings report within 48 hours: taxonomy health, duplicate events, governance gaps, and the three fixes with the highest data-trust payoff. No commitment.
Request the free audit