The Play closed-testing requirement
Twelve testers, fourteen days, and a production-access application at the end of it. Here is what each part of that actually asks for.
At least 12 testers, opted in at the same time
Not 12 installs, and not 12 people who tried it at some point during the month. Google counts testers who are opted into your closed track, and the count has to hold for the whole window.
For 14 continuous days
The window is continuous. If your count drops below the threshold partway through, you are not topping up a total — you are restarting a clock. This is the single most expensive thing to get wrong, because you only find out at the end.
Personal accounts only
The requirement applies to personal developer accounts created after 13 November 2023. Organization accounts verified with a D-U-N-S number are exempt, which is why some developers register a company rather than run the test at all.
What "continuously opted in" means
Opting in is an explicit action a tester takes with a specific Google account: they open your closed-track opt-in link, accept, and install from the Play Store while signed in as that account. Adding someone's email to your tester list does not opt them in. Until they follow the link, they are on a list and nothing more.
Staying opted in is passive, which is what makes it easy to lose. A tester who opts out, leaves the Google Group you manage the list through, or switches to a different account on their phone stops counting from that moment. They rarely mention it, and your Play Console does not shout about it either.
Uninstalling the app is not the same as opting out, but it may as well be for your purposes: a tester who has uninstalled generates no usage, and since 2026 usage is part of what is being judged.
How the rule has moved
Google has changed this requirement twice already. Treat the number in your own Play Console as authoritative and this page as context — including when the two disagree.
The engagement shift
The original rule was a counting exercise, and it was widely gamed: twelve accounts installed an app, never opened it, and the box was ticked. Rejections for thin engagement closed that door. The review now looks at whether your test produced something that resembles real use.
In practice that means testers who open the app on more than one day, reach more than the first screen, and leave the kind of trace a genuinely interested user would. You cannot fake your way past a reviewer looking at session data, and you cannot fix it retroactively — by the time the application is rejected, the fourteen days are already spent.
This is why we built a testing exchange around proof rather than matching: finding twelve people is the easy half.
Next: what actually goes wrong
Most failed runs are not failures of effort. They are a wrong link, a country restriction, or four testers who quietly drifted off in week two. Read the mistakes guide.
Running a closed test?
Get started