A user segment is a defined subset of users grouped by shared attributes or behaviors so you can analyze them separately and act differently for each group. Think “people who signed up this week and viewed pricing 3+ times” versus “returning blog readers on mobile.” Segments let you cut noise, compare patterns, and run targeted experiments across your traffic, not just the aggregate blur of all users.
Common ways to segment
- Behavioral: performed events, depth, recency, frequency, sequence.
- Lifecycle: new vs. returning (by session), activated, churn-risk.
- Acquisition: source/medium/campaign, landing pageview.
- Technographic: device, browser, connection.
- Journey: stages in a funnel leading to a conversion.
Why it matters
Segments turn averages into decisions. You can isolate where a funnel leaks for one audience, personalize onboarding, or compare conversion rates across channels. They’re the backbone for retention analysis and cohort studies.
Quick math
A segment is a filter, not a new metric. But you’ll often compute rates inside it:
Segment conversion rate = Conversions within segment / Users in segmentMini-example:
If 420 users clicked “Start Trial” out of 2,100 users who viewed pricing ≥3 times, the segment conversion rate is 20%.
| Segment filter (example) | Users matched |
|---|---|
| Viewed pricing ≥3 times in last 7 days | 2,100 |
| …and completed “Start Trial” | 420 |
Good practices
- Be explicit: define time windows, scopes (user vs. session) and inclusion logic (AND/OR).
- Avoid tiny Ns: small samples produce volatile rates; aggregate where sensible.
- Prevent leakage: don’t define segments using post-conversion data if you’re predicting conversion.
- Compare with a baseline: contrast to “All users” or a sibling segment.
- Name clearly:
pricing_3plus_last7d_mobilebeats “interested users.”