A segment is a subset of your analytics data that meets conditions you define, such as visitors on mobile, sessions that started from Google Ads, or users who bought more than once. In GA4, segments are built in explorations and let you compare that group with the rest of your traffic.
How segments work
You create segments inside explorations, GA4’s custom analysis area. There are three types, and choosing the right one changes the answer:
- User segments include all the activity, within the date range, of users who met the condition at some point. A user segment of “viewed the pricing page” includes those users’ other visits too.
- Session segments include only the sessions in which the condition was met.
- Event segments include only the individual events that meet the condition.
The scope of each dimension matters too: a user-scoped dimension such as first user source describes how someone originally arrived, while a session-scoped one describes each visit. Segments can also use sequences, for example “viewed a service page, then submitted the contact form”, which is useful for understanding routes to an enquiry.
Segments apply to data you have already collected, but only as far back as your data retention setting allows. Standard properties keep exploration data for two months unless you change it to fourteen. Large segmented queries can also be estimated rather than exact, through sampling.
Segments differ from an audience, which is defined in GA4’s admin, is built for acting on users rather than analysing history, and can be shared with Google Ads for targeting. You can turn a segment into an audience, but a new audience mostly gathers members from its creation onwards rather than from your full history.
Why it matters
Averages hide the useful detail. A site-wide conversion rate of 2% might be 4% on desktop and 1% on mobile, which points to a mobile form problem rather than a traffic problem. Segments let you split by device, channel, new or returning visitors, region, or any behaviour that matters to you, and see where the gap is.
UK businesses often want to compare London with the rest of the country, or a service area with everywhere else. GA4’s location data is estimated from IP addresses, so city figures are approximate, and it covers only visitors who accepted analytics cookies. Treat regional segments as directional rather than exact.
Common mistakes
- Using a user segment when you meant a session segment, and getting figures that include unrelated visits.
- Leaving data retention at two months, then finding last year’s segments return nothing.
- Building segments so narrow that GA4 hides rows through thresholding or the results rest on a few people.
- Confusing segments with audiences and expecting a new audience to contain everyone who ever met its conditions.
- Segmenting by things you cannot act on. A segment is only useful if it could change a decision.
How to act on it
Set data retention to fourteen months in GA4’s admin today, because the change does not recover data already lost. Then build an exploration with a small set of standard segments: mobile versus desktop, paid versus organic, new versus returning, and users who completed a key event. Keep it as a template and duplicate it for new questions.
When a segment reveals a gap, form a specific hypothesis, change one thing, and watch the same segment afterwards. Structuring analysis so it leads to decisions rather than more charts is part of my digital marketing strategy and consulting work.