Your analytics is only as good as your event names. A consistent taxonomy means your team — and the AI agents that read your metrics API — can ask questions without re-learning your data model. The goal is boring, predictable, queryable names.
Use snake_case for every event name: signup_click, plan_upgraded, checkout_completed. Reserve dots for namespacing vendored events, like adwords.conversion, so you never collide with your own events. Keep properties flat and typed: boolean for toggles, number for amounts, string for ids. Avoid reusing a property name for two different types — agents and humans alike will thank you.
Design funnels around verbs, not pages: start_checkout, add_payment_info, purchase. Because Tracabit stores eventParams per event, you can attach metadata like plan or value without changing the event name. That keeps your funnel definitions stable while your property vocabulary grows.
Document the taxonomy where your team already works — a markdown file in your repo or the Resources hub. Then enforce it in review: every new track() call should point to the documented name. Consistent events become the contract that makes agent-readable analytics reliable.
More guides
Privacy-first analytics implementation guide
Step-by-step setup for script tag integration, event taxonomy, and anonymous session handling without cookies.
Migrating from Google Analytics 4 to Tracabit
How to map GA4 event models and measurement IDs to Tracabit zero-cookie tracking in under 15 minutes.
Architecture: how Tracabit processes events
Deep dive into the Telemetry ingestion pipeline, sub-millisecond response times, and storage in Tero and ClickHouse.
Turn every signal into your next decision.
Start with a lightweight tracker. Grow into agent-readable product intelligence.
No cookies. No third-party ad tracking. Your telemetry stays yours.