Skip to content
Android closed testing

Closed testing with proofthe testing actually happened

12 testers for 14 days is the easy part — Reddit and Telegram will find you those. Whether any of them opened your app after day one is the part nobody can show you. Tavern verifies it every day, for every tester, and hands you the record.

One app per account at a time · No paid placement, ever

Finding testers is the easy half

Reddit threads, Telegram groups, Google Groups, and three competing apps will all hand you 12 installs. None of them can tell you whether those testers opened your app after day one — and since 2026 that is the part Google rejects applications over: installs with no real usage behind them.

Matching is free

Plenty of places will find you 12 people willing to opt in. We do not charge for the part that is already abundant.

Engagement is not

A screenshot from a group chat proves nothing. Neither does an install count. Both are what a thin closed test looks like from the outside.

So we build the record

Each tester submits evidence every day for 14 days, checked server-side, kept as a record you can actually read.

What actually gets checked

Every submission runs through the same server-side checks. Nothing rejects on a single signal — a legitimate tester whose gallery app rewrote a timestamp should never be called a cheat — but recycled evidence has nowhere to hide.

Perceptual hashing

Each screenshot is hashed on its pixels, so a re-encode, a resize, or stripped EXIF data changes nothing. Yesterday’s shot is recognised as yesterday’s shot — including one recycled from somebody else’s test.

A rotating daily target

You name three or four reachable screens. Each tester is asked for a different one each day, so an old screenshot shows the wrong screen and reuse fails by construction rather than by detection.

The server clock is the only trusted one

Device capture and queue times are sent along and corroborate, but a phone clock is user-settable. Skew, future times, and stale captures are flagged for review.

A day lock, enforced server-side

Evidence can only be filed for the current day of the run. Backdating and end-of-window batch uploads are not possible — and a late upload on bad signal still counts for the day it was taken.

Up to 3 shots, judged independently

Extra screenshots are free upside: duplicates among them are dropped, never held against the tester, and one clean shot on target counts the day. The verdict reads "day 6 counted, 2 of 3 images clean".

Session telemetry, if you use the SDK

Apps already running the SupportDock feedback SDK can report real sessions — how long, how far past the first screen — instead of a picture of one. Measurement rather than evidence of evidence.

What we do not claim

We prove your evidence is fresh, unique, and on the screen it was asked for. We do not promise that Google will approve your application. Nobody outside Google can, and any service telling you otherwise is selling something it does not own.

What you get is a clean record of real testing: no recycled proof, verified fresh every day. Review outcomes depend on your app, your listing, and Google's judgement — all three stay yours.

How a run works

1

List your app

Paste your Play Store listing URL and your closed-track opt-in link — they are different pages, and sending the wrong one is the most common way a test quietly fails. We check the opt-in link resolves before anyone is assigned to it.

2

Testers take seats

Other developers request seats. You approve each one, with their reliability and their own approval history in front of you when you decide.

3

The whole party starts together

Testing begins for everyone at once, so all 12 testers overlap in one continuous window rather than trickling in across it.

4

Evidence every day, checked every day

For 14 days each tester files evidence on the screen they were asked for. You watch the record fill in, day by day, tester by tester.

How the exchange is run

Ordered by need, never by payment

The queue puts the emptiest listing first. There is no promoted placement, no priority tier to buy, and no sort order for sale — the only thing that moves a listing up is needing testers more than the next one.

Testers are developers too

The people taking seats are shipping their own apps and know what a useful bug report looks like. Reciprocity is checked when seats are approved.

No rating trades. At all.

Soliciting or exchanging store ratings removes you from the exchange. There are no private tester-to-developer channels here, which is where that behaviour starts elsewhere.

Feedback goes on the record

Testers write what they saw alongside their evidence, and it lands in the test record where you read it. Optional by design — mandatory feedback is how you get fake feedback.

Understand the requirement first

Common questions