Product6 min read

Product Analytics

Instrument your product, track the right events, and analyze funnels, cohorts, and retention curves to make evidence-based product decisions.

Product AnalyticsAmplitudeFunnel AnalysisCohort AnalysisMetrics

skill: Product Analytics category: Product tags: [Product Analytics, Amplitude, Funnel Analysis, Cohort Analysis, Metrics] readingTime: 6

Good product analytics starts with a clear question, not a data dump. Define the metrics that matter for your business model before instrumenting. The failure mode is 200 events tracked but no one knows what they mean or how to query them.

Workflow

Define your North Star MetricA single North Star Metric aligns the whole company - it represents the core value users get from the product and correlates with long-term revenue.
Design your event taxonomyInconsistent event naming ('user_signed_up' vs 'Signup Completed' vs 'account_created') makes analytics queries brittle and reports unreliable.
Analyze funnels for conversion dropsFunnel analysis shows where users drop off between steps - the step with the highest drop-off is the highest leverage optimization target.
Run cohort retention analysisCohort analysis separates users by when they joined and tracks retention over time - revealing whether recent cohorts are improving or degrading.
Set up dashboards with actionable metricsDashboards nobody checks are vanity projects. Good dashboards surface the 5-10 numbers that tell you whether the product is healthy.

Output Format

Analytics system: North Star Metric, event taxonomy, funnel + cohort dashboards, and weekly health report.

Analytics Stack:
  Events: [Object] [Action] naming, tracking plan doc
  Metrics: NSM + 4 input metrics on daily dashboard
  Funnels: activation, checkout, upgrade flows
  Cohorts: weekly retention heatmap
  Alerts: Slack on >20% WoW drop in any key metric

Examples

Input

Output

Edge Cases & Gotchas

Event tracking fires on errors - If a checkout completion event fires even when payment fails, your conversion metric is inflated. Add server-side validation to analytics events for money-sensitive flows.
Segment size vs. significance - A feature with 0.5% adoption that serves enterprise customers may be more valuable than one with 20% adoption. Segment by user type before concluding a feature isn't working.
Privacy and GDPR compliance - GDPR requires opt-in analytics in EU. Use consent management (Osano, OneTrust) before initializing analytics SDKs. Anonymize user IDs in analytics exports.
Session ID vs. user ID tracking - Anonymous sessions before login can't be linked to the user account post-login without an alias() call. Implement identify() + alias() to maintain a continuous user journey.

When to Break These Rules

Use SQL + dbt on your data warehouse (BigQuery, Snowflake) instead of Amplitude when analyses require joining product data with revenue, support, or CRM data.
Use Posthog (open source, self-hosted) instead of Amplitude when data privacy regulations or budget prevent using cloud analytics tools.

Reference Files

https://amplitude.com/blog/product-analytics
https://www.lennysnewsletter.com/p/north-star-metric