Segments and segment overlap

A segment is a subset of your data defined by conditions. Create one from the Variables panel (+ next to Segments) or by right-clicking on the canvas. Besides suggested segments, custom ones come in three types:

• User segment: all events from users who match. Users who purchased includes their browsing from before the purchase.
• Session segment: all events from sessions that match. Sessions from Paid Search keeps those visits, not the same users' organic visits.
• Event segment: only the matching events, e.g. purchase events with value > 100.

Pick the scope by asking: everything about these people, everything in these visits, or only these events?

Conditions use dimensions and metrics with operators (equals, contains, regex, greater than, is one of), grouped in Include and Exclude blocks with AND/OR. Each condition group has its own scoping: Across all sessions (default for user segments), Within the same session or Within the same event. "Viewed pricing AND purchased" means something different across all sessions than within one session.

A user segment has a Build an audience checkbox that creates a matching audience for Google Ads; the reverse is impossible.

Filters vs segments: a tab filter removes rows from the output; a segment changes the population before metrics are computed, and only segments can express exclusions and sequences. A user segment of German users also brings their sessions from abroad; a filter Country = DE would not.

Segment overlap is the technique that draws up to three segments as a Venn diagram, with a table of every combination and the metrics you choose. Use it to check whether two groups are really the same people (mobile users vs purchasers). Right-click a region and choose Create segment from selection to turn an intersection into a reusable segment. Users are counted even for session and event segments, so mixing scopes is legal but easy to misread.

Segments are retroactive, which audiences are not, and up to four apply per tab.