Glossary
Event data model
The structured format — event name, timestamp, user ID, and properties — that a single tracked user action is recorded in.
Also called: event schema, event model
An event data model defines the structure every tracked action in a product or website is recorded in: typically an event name, a timestamp, a user or device identifier, and a set of properties describing what happened, such as the page viewed or the item added to a cart. Every clickstream a site or app collects is, underneath, a stream of records following this shape.
Most event data models share a common skeleton — event_name, timestamp, user_id, properties {...} — but teams differ in how strict that shape is. Some enforce a fixed event taxonomy of allowed event names and property types before data lands in the warehouse; others let any event through and clean it up downstream. This differs from a relational transaction record, which describes a completed business object like an order, rather than a raw, timestamped action a person took.
A well-designed event data model matters because most downstream analysis — sessionization, funnel counts, user journey analytics — depends on consistent event names and properties. An "add_to_cart" event logged three different ways across platforms breaks every report built on top of it. The most common pitfall is instrumenting events ad hoc as questions arise, rather than planning the model up front, producing a taxonomy that is inconsistent and expensive to reconcile later.
Last reviewed September 22, 2026