Event Taxonomy
An event taxonomy is a governed naming and classification system for product events and properties, designed to keep behavioral data consistent.
Download Markdown (.md)Soroush’s Product Wiki · https://wiki.soroush.uk/topic/event-taxonomy/
01What it means
An event taxonomy defines how meaningful product occurrences are represented in analytics data. It establishes event and property names, grammar, casing, entities, shared context, allowed values, ownership, and lifecycle rules. The taxonomy is the vocabulary; a tracking plan lists the specific events to implement.[1][2][3]
A taxonomy should serve defined decisions and remain versioned; collecting every possible event increases privacy risk and analytical noise.
02Why it matters
Without shared rules, equivalent actions appear under several names, details become duplicate events, and analysts cannot compare journeys confidently. A clear taxonomy improves discovery, implementation consistency, governance, and trust while limiting a sprawling catalog of ambiguous clicks.
03When to use it
Create a taxonomy before substantial instrumentation, when multiple products or teams share data, or when an existing event catalog has become inconsistent. Apply it to new events and migration work, but avoid renaming historical concepts casually because downstream analyses may depend on them.
04How it works
- Identify core entities and user actions from priority product questions and journeys.
- Choose a readable naming pattern, grammatical tense, casing convention, and rules for events versus properties.
- Define shared properties such as identifiers, platform, version, source, and timestamps with permitted values.
- Document ownership, approval, deprecation, versioning, and privacy rules.
- Validate incoming data against the taxonomy and review exceptions before expanding the vocabulary.
05A practical example
A commerce team standardizes past-tense object-action names such as “Product Viewed,” “Cart Updated,” and “Order Completed.” Channel, device, currency, and item count remain properties rather than spawning separate events. Deprecated names are mapped and monitored during a migration period.
06Where it breaks
A taxonomy cannot ensure correct firing logic, complete coverage, or useful analysis by itself. Overly rigid governance slows delivery, while overly generic events shift confusion into properties. Historical data may remain fragmented after conventions improve.
Common mistakes
- Encoding every variation as a new event instead of a property.
- Using vague names such as “Clicked” without the object or business meaning.
- Changing conventions without migration, ownership, or deprecation rules.
07Sources
- Free Website Event Taxonomy GeneratorAmplitude · Product analytics taxonomy guidanceFree Website Event Taxonomy Generator
- The Foundation for Great Analytics Is a Great TaxonomyAmplitude · Industry practice articleThe Foundation for Great Analytics Is a Great Taxonomy
- Official Events and PropertiesAmplitude Docs · Official governance documentationOfficial Events and Properties