How GA4 decides someone is a new user
When activity arrives, GA4 looks for an identifier it already knows. On the web that is normally the client ID stored in a first-party cookie; in an app it is the app instance identifier; on a site with logged-in users it can be your own User-ID. If no earlier record exists for that identifier, GA4 records a first-visit event and counts the person as a new user.
Read that sentence carefully, because it is the whole subject. New does not mean a new human being. It means an identifier GA4 has not seen before. The same person becomes new again whenever the identifier is gone — a cleared browser, a different browser, a phone instead of a laptop, private browsing, a device that limits cookie lifetimes, or a visit where tracking consent was declined.
Why new users matter
It is the cleanest signal you have of whether your marketing is reaching people it has not reached before. A campaign that lifts sessions but not new users is mostly bringing back the audience you already had, which is fine for remarketing and disappointing for an awareness budget.
Split by channel, it tells you which sources do the introducing and which do the closing. Search and social often carry the first meeting; direct visits and email often carry the return. Judged that way, the number helps you decide where the top of the funnel actually is rather than which channel to trim.
Common mistakes with new users
The first is reading the count as unique people. It is always inflated, and by how much depends on your audience’s habits rather than on anything you control. Where visitors use several devices, or where a browser aggressively expires storage, the same person can appear as new on many occasions.
The second is expecting new users and sessions to reconcile with any other system. Ad platforms, your CRM and GA4 all identify people differently, so the same week will produce different figures everywhere. Chasing an exact match wastes days.
The third is comparing across a change you made. Adding a consent banner, moving to a new domain, breaking cross-domain tracking or reinstalling a tag will all create a bulge of apparently new users that has nothing to do with marketing. Record the dates of technical changes next to your traffic charts, or someone will eventually present that bulge as growth.
How to act on it
Use it as a direction rather than a headcount. Watch the trend within your own account, compare channels against each other over the same period, and pay attention to the ratio of new to returning rather than the raw figure.
Improve the accuracy where you reasonably can: keep cookies first-party, set up cross-domain measurement if visitors move between your sites, and send a User-ID for logged-in customers so their devices join up. Then judge campaigns on what those new users went on to do — enquiries, bookings, purchases — because a rise in first-time visitors that produces nothing is a reach story, not a result. Where the counts look strange rather than merely soft, a tracking review is the right next step.