Metrics
New Users vs Returning Users in GA4
Updated 2026-09-145 min read
Short answer
New users are people who logged their first open in the period you selected. Returning users had at least one earlier session. Both are counted inside the date range, so the split changes when the range changes — and neither tells you on its own whether this month's new users will come back.
This split is the difference between growth and retention. An app can post a healthy active user count while quietly failing, if nearly all of those people are new and nearly none of them come back.
What each side means
GA4 counts a new user when someone logs first_open — the first launch after installing — in the date range. A returning user is someone who had at least one earlier session, whether or not that session was engaged. Active users is a third, stricter count: it only includes people who actually engaged.
New users: 400 Returning users: 100 Reads as: strong acquisition, weak retention. Most of what you spent on getting people here left again.
The same 500 split the other way — 100 new, 400 returning — describes a completely different product, even though the headline number is identical.
Why the two do not always add up neatly
People expect new plus returning to equal active users exactly. Often it is close, but several things get in the way.
- New users is counted from first_open whether or not the person went on to engage, while active users requires engagement — so on a quiet day new users can even exceed active users
- First-time status is tied to how a user is identified, and identity can be lost — a reinstall or a new device looks like a new person
- Privacy thresholds can withhold small figures, so a breakdown may not sum to its own total
Note: Treat these as two informative measures rather than two halves that must reconcile to the decimal. If you need them to balance exactly, you are asking the data for a precision it does not have.
The reinstall problem
This one catches everybody eventually. Someone who deletes your app and installs it again, or moves to a new phone, may be recorded as a new user despite being an existing customer. Your new user count therefore includes some people who are not new to you at all.
Careful: Do not treat new users as a count of unique humans who have never heard of you. It is a count of first-time-seen identities, which is a related but different thing.
How to read the split
| Pattern | Usually means |
|---|---|
| New high, returning low | Acquisition works, the product does not hold people |
| New low, returning high | Loyal base, but growth has stalled |
| Both rising | Healthy growth with retention keeping pace |
| New spike, returning flat | A campaign or feature landed, but changed nothing lasting |
Checking whether people really come back
Returning users answers one question: did anyone who had used the app before show up in this window. It cannot tell you whether the people you acquired this month will stay, because long-standing regulars can keep the number healthy while every new cohort leaves. Four checks close that gap.
Follow a cohort instead of a window
Look at D1, D7 and D30 for people who started in the same week. The group is fixed on day zero, so a rise or fall is about those people, not about whoever happened to be active.
Translate DAU/MAU into days
Average daily active users divided by monthly active users, multiplied by 30, is roughly how many days a month the typical monthly user shows up. If that figure falls while both counts grow, new people use the app less often.
Make a return mean the core action
In a GA4 cohort exploration, set the return criterion to your core event — a completed workout, an exported file — instead of any activity. Opening the app to dismiss a notification is not coming back.
Rule out identity resets
Reinstalls and new phones turn returning people into new ones. If people sign in, setting a User-ID lets GA4 recognise them across installs; comparing store redownloads with new users shows how large the effect is.
This week: 1,200 active users, 300 of them returning (25%) Cohort that started 5 weeks ago: 900 new users D7 = 11% (99 users) D30 = 4% (36 users) Returning users are propped up by long-time regulars. New cohorts are not staying.
Careful: A jump in returning users right after a push or email campaign measures the campaign's reach, not a habit. Check whether the same people come back again the week after.
How this looks in Appnaly
Appnaly reports new users alongside the completed windows rather than as a standalone figure, because a new user count means very little without the active total beside it. Four hundred new users is excellent or alarming depending entirely on what the rest of the number looks like.
If you want to go further than the split, retention follows a specific group forward in time instead of describing a single window. The retention article covers how D1, D7 and D30 are read, and the utility app article covers apps that people open only when a need comes up.