Data Driven
04

Instrumentation: How Startups Actually Capture Data

You cannot analyze what you do not capture. Event tracking, analytics tools, and a clean data warehouse are the foundation of every serious data culture.

#engineering

Data does not magically exist. Engineers and product people have to instrument it.

Common tools

  • PostHog, Mixpanel, Amplitude , product analytics
  • Google Analytics, Plausible , web traffic
  • Segment, RudderStack , event pipelines
  • Snowflake, BigQuery, Postgres , data warehouses
  • Metabase, Looker, Hex , dashboards and exploration

What good instrumentation looks like

  • A clear event taxonomy (consistent names)
  • User identity stitched across devices
  • Server-side events for anything that touches money
  • Privacy and consent handled correctly
Bad data is worse than no data , it produces confident, wrong decisions.