Glossary · Business Analytics
Event Taxonomy
also: tracking plan · event spec · analytics schema · event naming convention
Definition
Event taxonomy is the schema design discipline applied to product analytics: object-action-context naming, entity-event separation, PII boundaries baked into the schema, versioning rules, and validation at write time. A drifting taxonomy is the dominant cause of analytics debt at scale.
Event taxonomy treats product analytics as a data engineering problem. The dominant convention pairs an object (User, Order, Subscription) with an action (Created, Viewed, Cancelled) and an optional context modifier. The hardest design choices: where to draw the PII boundary (taxonomy enforces it, not application code); how to version events (additive properties are safe, semantic renames require new event names plus deprecation windows); and what to validate at write time (SDK-level schema validation through tools like Snowplow's Iglu registry catches drift before it lands in the warehouse). Most analytics rework projects exist because the original taxonomy was implicit rather than designed.
Essays on this concept
- Business Analytics
Event Taxonomy Design as Data Engineering
Event taxonomies are schema problems, not marketing problems. Teams that treat tracking plans as living documents (with versioning, validation, and PII boundaries) avoid the drift that quietly costs everyone else.
- Business Analytics
Identity Resolution in a Cookieless World: A Probabilistic Reality
The cookie was always probabilistic. Cookieless makes the probability legible. Operators who treat new identifiers as deterministic will misattribute spend and contaminate downstream measurement.
- Business Analytics
Server-Side Tagging Beyond Compliance: The Operational Case
Privacy compliance is the entry point for server-side tagging. The operational case is broader: latency, ad-blocker resilience, data quality, and the cost model of running an event router at production scale.
- Business Analytics
Cohort Analysis at the Action-Set Level (Not User-Level)
Sign-up-month cohorts confuse arrival with behavior. Action-set cohorts predict retention earlier and more honestly, at the cost of an event taxonomy, materialized views, and resolved identity discipline.
Related concepts