GA4 Custom Dimensions vs. Metrics: How to Choose

In GA4, dimensions describe or categorize data, while metrics quantify it. A dimension might identify a form type, plan tier, or content group; a metric might count a form submission or record a numeric value. Choosing the right type before implementation makes reports easier to build and compare.

Start with the question

Ask whether the value names a category or measures an amount. “signup_source” with values such as webinar, organic search, and partner is a dimension. “items_added” or “score_value” may be a metric if it is numeric and meaningful to aggregate. Do not turn every parameter into a report field: collect only data that supports a defined decision.

Choose the appropriate scope

Custom dimensions can be event-scoped, user-scoped, or item-scoped for ecommerce data. Scope describes where the value applies. An event-scoped dimension belongs to a particular event; a user-scoped value describes a user property that can persist across events. Item-scoped values describe product items. A mismatched scope can make a report appear to assign one value to the wrong activity.

Register only useful parameters

When a custom event parameter needs to appear in GA4 reporting interfaces, register it as a custom definition with the correct scope and name. Registration does not create the event or repair its payload: the tag or application must already send the value consistently. Standard dimensions and metrics do not need duplicate custom definitions.

• Use a stable parameter name and a controlled set of dimension values.
• Send numeric values as numbers when they are meant to be measured.
• Avoid high-cardinality dimensions that produce many one-off values.
• Keep personal or sensitive information out of event parameters.

Validate before relying on a report

Check the incoming event and parameter in a debugging view or network request, then confirm the registered definition appears in the intended reporting surface. Test representative values and confirm the scope matches the question. Custom definitions are not a retroactive way to make every prior event available in every report, and some reports have processing or cardinality limits.

A concise measurement plan should record each parameter, its meaning, scope, allowed values, and the report or decision it supports. That small data dictionary prevents dimensions and metrics from drifting as teams add events.