Production Access
You have 12 testers and Google still said no
You met the requirement. Twelve testers, fourteen days, counter satisfied. And Google still told you to keep testing. This is not a bug and it is not bad luck: meeting the testing criteria is what lets you apply, not what gets you approved. Google's own documentation says it outright — if your app isn't ready to be published, you may be required to continue testing.
The two decisions people confuse
Production access
Permission for your account to publish to production at all. This is the gate the 12 testers and 14 days unlock, and it is decided partly on the questionnaire you fill in.
Release review
Whether a specific build and store listing are allowed to go live. Policy, content rating, Data Safety accuracy and listing consistency are judged here. Being granted production access does not exempt any future release from this.
Developers routinely blame "the 12 testers thing" for what is actually a release-review problem, then re-run a testing cycle that was never the issue. Read the wording of what Google sent you and work out which of the two you are actually facing before you do anything.
Why a cycle that satisfied the counter still reads as weak
| What Google sees | What actually happened | What to do |
|---|---|---|
| Testers enrolled, almost no usage | People accepted the opt-in link to help you out and never opened the app again. | Give testers specific tasks, not a link. Ask for one core action each and collect the answers in writing. |
| Feedback summary is generic | You collected "works fine" from a group chat and had nothing concrete to summarise. | Collect issues with device, OS version, steps and expected vs actual. Even five real notes beat fifty thumbs-ups. |
| No described change from testing | Nothing was fixed, because nothing was found, because nothing was really tested. | Ship at least one improvement traceable to the cycle — and if genuinely nothing needed fixing, document what you tested and how you verified it. |
| Listing and app disagree | Screenshots or description promise features testers never saw. | Align the listing with the build before applying. This is release-review territory and it bites at the same moment. |
| Answers are vague or templated | The questionnaire was filled in as fast as possible to get past it. | Answer from your actual cycle. Specifics are the entire signal here. |
Google publishes no minimum number of sessions, actions or feedback submissions, and no scoring formula. The rows above are what we observe across cycles, not a documented rubric.
The counting rule people get wrong
Google counts the fourteen days per tester, not as a group streak. In Google's wording, testers who opt in, test for fewer than 14 days and then opt out are not counted — and if they opt back in, the days still have to be consecutive to count.
The practical consequence catches people out: a tester leaving on day 2 costs you nothing, because a replacement still has time to complete their own 14 days before you apply. A tester leaving on day 13 cannot be replaced in time, no matter how fast you react. This is why checking enrollment daily matters and "topping up at the end" does not work.
What to do after a rejection, in order
Read what they actually said
Separate "keep testing" from a policy or listing objection. They need completely different responses.
Do not resubmit immediately
An identical resubmission with no new evidence spends time you can't get back.
Go get real feedback
Go back to your testers with specific questions: what worked, what was unclear, what broke. Written answers, with device and OS.
Fix something and record it
Even one meaningful fix gives you a truthful answer to "what changed because of testing".
Rewrite the questionnaire
Answer from the cycle you actually ran. Concrete beats confident.
Related: what a real closed testing report contains — including the exact questionnaire sections — and the requirement explained. If testers can't get in at all, that's a different problem: the opt-in link troubleshooting guide. Before you resubmit, read the questionnaire question by question.
Cycle didn't hold up?
Run one that produces evidence, not just a counter.
Real testers for the full 14 days, tasks that generate usable feedback, and a QA report you can answer the production questionnaire from. Plans from $180, worldwide.
Start your testing cycle