Data View and Component Configuration
What you must understand
A data view is where raw schema fields become reportable components: dimensions, metrics, segments, and date ranges. Know the component settings that shape analysis: dimension behavior (e.g., whether values are treated as low-cardinality or high-cardinality), attribution models on metrics, and the session settings (session timeout, whether a new session starts on a campaign or event).
Understand how metrics are built-count, sum, or derived-and how attribution (last touch, first touch, linear, participation) changes metric results. Know that a data view has a default attribution model that can be overridden per metric. Understand persistence settings on dimensions (e.g., how eVar-like persistence works) and how they affect segmentation.
Traps to watch
- Attribution is set on the metric, not the dimension; questions often invert this.
- Session timeout is a data view setting; changing it later affects historical sessionization.
- Component naming and descriptions matter for governance and for avoiding duplicate components.
- A dimension's persistence and allocation settings change segment results, not raw counts.
Worked example
If a visitor sees Campaign A then Campaign B before converting, a last-touch attribution metric credits B; a first-touch metric credits A; a linear metric splits credit. The underlying event data is identical-only the metric's attribution setting changes the number. Choose the attribution model that matches the business question.