Testing a Permission Priming Screen Before the Prompt

Heavy denials at the system prompt, support tickets asking what a permission is for, or trust concerns in surveys suggest users do not see why the app is asking, which is the gap a pre-permission screen tries to fill. If the explanation works, acceptance rises among users who pass it and more than covers the drop-off on that screen, so grants rise across all assigned users. If denials come from asking at the wrong moment, an explanation will not fix that; handle timing in a separate test. Check store rules before launch, make the screen clearly distinct from the system prompt, and leave a way to continue without granting permission.

Net permission grant rate

Compare permissions granted per eligible assigned user over the same follow-up period.

Also track

• System prompt acceptance rateCheck acceptance among users who reach the prompt.
• Pre-prompt continuation rateIdentify drop-off at the explanatory step, which appears only in the variant.
• Permission denial rateUse the platform's available status information; do not assume every denial is permanent.
• Seven-day retentionMake sure retention does not fall.

• ExplanationAdd one fixed explanation before the existing system permission prompt.
• TriggerKeep eligibility and the permission trigger unchanged.
• Drop-offInclude visitors who defer or leave before the system prompt.
• DeferralPredefine the “Not now” path and later re-prompt policy.
• ExperienceReview eligible new and experienced users separately.

1. 1Do not declare success from a higher system prompt acceptance rate, because in the variant only users who pass the explanation reach the prompt. Judge it by the net permission grant rate across all eligible assigned users, keep users who defer in the denominator, and confirm retention does not fall.
2. 2Read this result as specific to this explanation copy and the current moment when permission is requested. Review new and experienced users separately, track denials with the platform's status information, and do not assume every denial is permanent.
3. 3Users who defer and grant permission later can shift the result late, so with a small gap check whether the follow-up period is long enough to capture them. Put each arm's eligible users and net grants into the A/B test significance calculator to see whether the gap stands out from chance, and keep the “Not now” path unchanged throughout the test.