GA4 segments: how to create them and what they're for

A segment in GA4 is a filtered subset of your data, and GA4 offers three kinds: user segments, session segments, and event segments. They differ in what they return, not just in what they filter.

The critical thing most people learn the hard way: segments only exist inside Explorations. You can't apply a segment to a standard report. If you have been hunting for the segment button in Reports, it isn't hidden. It isn't there.

Segments also aren't audiences, even though the builder looks nearly identical. A segment analyses the past. An audience collects users going forward and can be pushed to Google Ads. Confusing the two costs people weeks.

This article covers how to build each segment type, what each one actually returns, the limits nobody mentions, and when you should be building an audience instead.

The three segment types, and why the difference matters

Pick the wrong scope and you get a technically correct answer to a question you didn't ask.

User segment. Selects users who meet the condition, and returns everything those users did, across every session in the date range.

Build a user segment for "purchased something" and you get all their sessions, including the ones where they only read the blog. That's the point: it lets you study behaviour that led up to a conversion.

Session segment. Selects only the sessions that meet the condition.

The same "purchased" condition as a session segment returns only the sessions containing a purchase. The blog-reading session from the same user is excluded.

Event segment. Selects individual events matching the condition, not the sessions or users around them.

Narrow and rarely what you want, but useful when you're debugging a specific event's parameters.

The one example that makes it click

You want to know which landing pages lead to purchases.

• With a session segment on purchase, you see the landing page of the session where the purchase happened. Often that's the cart page or a branded search. It tells you almost nothing about what created the demand.
• With a user segment on purchase, you see every landing page those buyers ever entered on, including the article that first brought them in three weeks earlier.

Same underlying data. Two completely different answers. The second one is usually the question you actually meant.

Are the events behind your segments even firing correctly? The Free GTM Audit checks the tracking layer that feeds them. Enter a URL, get a report in about 10 minutes. No signup.

How to build a segment, step by step

1. Open Explore in the left navigation and create a new exploration (Blank works).
2. In the Variables column, click the + next to Segments.
3. Choose User, Session, or Event segment. This is the decision that matters most, and GA4 asks it first for a reason.
4. Add your conditions. Common ones: Event name exactly matches purchase, Session default channel group exactly matches Organic Search, Device category exactly matches mobile.
5. Name the segment something a colleague would understand. "Mobile buyers, organic, last 30d" beats "Segment 1".
6. Save and apply. Drag it onto the Segment comparisons area of your exploration.

The segment now filters that exploration. It doesn't affect standard reports, and it isn't shared with anyone else unless you save it to the property scope.

Conditions vs sequences vs exclusions

Three tools sit inside the segment builder and they do different jobs.

• Conditions match on attributes. "Users who viewed a product." Order doesn't matter.
• Sequences match on order. "Users who viewed a product, then added to cart, then didn't purchase." Use these when the order is the insight.
• Exclusions remove users or sessions. "Users who purchased" removed from a funnel gives you the people who dropped out.

Sequences are where segments earn their keep. A condition tells you who did something. A sequence tells you what they did before they gave up.

Segments vs audiences: the distinction that costs people weeks

The builders look almost identical. The behaviour isn't remotely the same.

	Segment	Audience
Where it lives	Explorations only	Admin, property level
Retroactive?	Yes. Analyses existing data	No. Starts collecting from creation day
Can export to Google Ads?	No	Yes
Purpose	Analysis. Understanding what happened	Activation. Targeting people going forward
Membership	Recalculated each time you run it	Users accumulate over time

The one that hurts: audiences aren't retroactive. Create an audience for "abandoned cart" today, and it contains nobody. It starts filling tomorrow. People routinely build an audience, see zero users, assume it's broken, and rebuild it three times before someone explains this.

If you want to analyse cart abandoners from last month, that's a segment. If you want to retarget them in Google Ads, that's an audience, and you needed to create it last month.

Marta, a marketing manager at a Polish e-commerce company, spent two weeks debugging an "empty" audience for high-value customers before we found the problem: nothing was wrong. The audience had been created on Monday and she was looking at it on Tuesday. It was working exactly as designed, and the design is unintuitive.

The limits nobody mentions

Explorations are sampled. Above roughly 10 million events in your date range, GA4 samples. Your segment then returns an approximation, and the sampling notice is a small icon in the top right that's easy to miss. On a large property, always check it before you present a number to anyone. If you need unsampled data at that scale, the answer is the BigQuery export, not a better segment.

Segment limits. You get 10 segments per exploration, and 100 saved segments per property. Enough for most work, but worth knowing before you build a library.

Segments don't exist in standard reports. Standard reports offer comparisons, which look similar and are much more limited: fewer conditions, no sequences, no scope choice. If your analysis needs a sequence, you're in an Exploration whether you like it or not.

A segment is only as good as the events under it. A segment on purchase is meaningless if the purchase event fires twice, or fires on the cart page, or stopped firing after the last deploy. The segment will confidently return a wrong answer, because it has no way of knowing the data is wrong. Every analytical layer inherits the defects of the tracking below it.

Turning a segment into an audience

Once a segment tells you something useful, you usually want to act on it. That means an audience.

GA4 gives you a shortcut. Inside an exploration, right-click a segment and choose Build audience. It carries your conditions across, so you don't rebuild them by hand.

Three things to know before you do it:

• The clock starts now. The audience begins collecting members from the moment you save it. Everything the segment showed you about the past does not transfer. You have the insight, not the people.
• Google Ads needs the link. An audience only reaches Google Ads if the GA4 property is linked to the Ads account and the audience is set to share. Check this before you plan a campaign around it.
• Membership duration matters. The default is 30 days. For a considered purchase with a 60-day cycle, a 30-day window quietly drops people who were still deciding.

The practical implication: if you can already predict which groups you'll want to target, create those audiences now, even before you need them. An audience created today is worth something in a month. An audience created in a month is worth nothing today. This is the single most common piece of advice I give that people ignore, and then regret.

Four segments worth building today

Practical ones, in order of how often I use them.

1. Converters vs non-converters. A user segment on your key event, and its exclusion. Compare landing pages, channels, and devices between the two. This is the fastest way to find where value actually enters.
2. Cart abandoners. A sequence: add_to_cart followed by no purchase. Then look at what page they were on when they gave up.
3. Engaged non-buyers. Users with more than three sessions and no purchase. This is your remarketing goldmine, and it's also your pricing objection sample.
4. Mobile vs desktop buyers. A user segment split by device. If mobile converts far worse on the same traffic, the problem is usually the checkout, not the audience.

Start with number 1. It answers a real question, it takes five minutes, and it will usually surprise you.

Frequently asked questions

What is a segment in GA4? A segment is a filtered subset of your data, used inside Explorations. GA4 has three types: user segments (everything those users did), session segments (only matching sessions), and event segments (only matching events).

What's the difference between a segment and an audience in GA4? A segment analyses existing data retroactively and lives only in Explorations. An audience collects users from the day you create it, can't look backwards, and can be exported to Google Ads for targeting.

Why can't I apply a segment to a standard GA4 report? Because segments only work in Explorations. Standard reports offer comparisons instead, which are more limited: no sequences and no choice of scope.

Are GA4 audiences retroactive? No. An audience starts empty and fills with users who qualify from its creation date onward. If you need to analyse a group from last month, use a segment.

How many segments can I have in GA4? Up to 10 segments per exploration and 100 saved segments per property.

Can I turn a GA4 segment into an audience? Yes. Right-click the segment inside an exploration and choose Build audience. The conditions carry over, but the membership does not: the audience starts empty and collects users from that day forward.

Why do my segment numbers look wrong? Two usual causes. First, sampling: Explorations sample above roughly 10 million events, so the number is an approximation. Second, scope: a user segment returns everything those users did, not only the sessions that matched, which is often mistaken for an error when it's the intended behaviour.

Next steps

Segments are the most useful part of GA4 that most people never open, and the scope choice is where nearly all the confusion lives. User scope returns everything those people did. Session scope returns only the matching visits. Pick the wrong one and you get a precise answer to the wrong question.

Build the converters vs non-converters pair first. Compare their landing pages and channels. Then check the sampling notice before you quote any number from it.

And before you trust any of it, make sure the events underneath are real. A segment can't tell you that your purchase event fires twice, it will simply report a number that's twice as confident as it should be. The Free GTM Audit checks that layer. If you want it verified every month rather than the next time something looks strange, that's what independent GA4 monitoring is for. €150/month, written report, no calls, cancel anytime. Full pricing here.

---

Sources and further reading

• Build and edit segments in GA4 (Google)
• Audiences in GA4 (Google)
• Data sampling in GA4 (Google)