✅
GuideOctober 21, 2026

14-Day Closed Testing Checklist: 30 Pre-Production Checks

A day-by-day checklist of 30 checks for the closed testing phase — setup, first sessions, stability metrics, evidence for the production request — so day 14 becomes a formality.

AI Summary & Key Takeaways

  • ✔The checklist splits the 14 days into four stages: 8 setup checks before day 1, 6 first-session checks on days 1–3, 7 stability checks on days 4–9, 6 evidence checks on days 10–14, and 4 items for the production request itself.
  • ✔The most expensive silent failure is a broken opt-in link — test it from a logged-out browser before inviting testers.
  • ✔Target 99.5%+ crash-free sessions, zero ANRs, and daily opens from every tester; twelve installs that go cold read as a failed test to Google.
  • ✔Keep evidence exportable: opt-in counts, active devices and daily engagement screenshots support the production questionnaire.
  • ✔The production questionnaire must be answered honestly and consistently with your Console data — that is what manual review checks.
  • ✔Recruit 15–18 testers for 12 slots: real people on real devices who were told exactly what participation means — flaky friends and emulator farms poison the engagement data.
  • ✔One daily nudge message with a concrete task is the highest-leverage habit of the phase; templates for days 1, 3, 5, 7, 10 and 13 are in the article.
  • ✔The track’s numbers (opt-in rate, daily actives, uninstalls, crash-free sessions, device spread) announce failures days early — a symptom-to-fix table shows how to react within a day.

Fourteen days sounds like a long time until you are living inside them. Between recruiting testers, watching Android vitals, and wondering whether Google's algorithms see what you see, it is easy to reach day 14 unsure whether you are actually ready to request production access.

This checklist breaks the closed testing phase into 30 concrete checks across four stages: setup, first sessions, mid-phase stability, and the production request itself. Print it, paste it into your issue tracker, or work through it day by day — the numbering follows a realistic 14-day calendar for a small team.

Two ways to run it: solo, where you own every line and the checklist doubles as your daily to-do list, or as a team, where each stage gets an owner — someone watches testers, someone watches vitals, someone owns the store listing. Either way, review the relevant block every morning; the checklist is ordered so that the things you must fix today appear before the things you must fix tomorrow. The sections between the numbered items explain the "why" behind the checks that most often decide pass or fail.


Before Day 1: Setup Checks (1–8)#

Everything on this list must be true before your first tester opts in. Fixing store-listing problems mid-phase resets momentum and burns testing days.

  • App bundle uploaded to the closed track — not the internal track. Verify under Testing → Closed testing that the latest .aab is the active release.
  • Country availability set deliberately. If a tester's country is missing from the track, the app is invisible to them and your opt-in rate quietly bleeds.
  • Store listing complete: app name, short description (80 chars), full description, feature graphic, screenshots for phone (and tablet if supported). A bare listing lowers tester trust and opt-in rates.
  • Content rating questionnaire completed — an unrated app cannot publish to any track.
  • Data safety form filled accurately. Every permission and SDK that collects data should be reflected here; mismatches between the form and the app are a leading cause of review delays later.
  • Privacy policy URL live and reachable, matching the data safety form.
  • Target API level current for 2026 requirements (API 35+ for new apps), with no deprecated permission flags you cannot justify.
  • 12+ testers recruited with real devices — friends with spare phones, community testers, or a managed service (15 testers for 14 days is available from $47). Confirm they can all receive the opt-in link.

Open your own opt-in link from a logged-out browser and from a spare device. A broken or confusing join flow is the single most expensive silent failure: testers bounce, opt-ins never register, and you discover it at day 10.


Recruiting Testers Who Actually Stay#

Twelve is the minimum, but the composition of your twelve decides whether day 14 is a formality or a cliff. Google's engagement data cannot tell a real tester from a fake one by name — it tells them apart by behavior: installs that open daily versus installs that vanish. Recruit accordingly.

Who Counts, and Who Quietly Hurts You#

  • Counts well: real people with their own devices, in countries available on your track, willing to open the app once a day for two weeks.
  • Counts, but flaky: friends who mean well and forget. Recruit more than twelve precisely because some will fade; fifteen to eighteen recruits for twelve slots is a sane buffer.
  • Hurts you silently: emulator farms, VPN-warped devices, tester pools that install from one account on five phones, or anyone who uninstalls after the first day. Uninstall spikes and device clustering are exactly the patterns Google's anti-abuse systems look for — a "cheap tester" package that behaves like this is worse than no testers at all.

Where Real Testers Come From#

  • Your own circles, with obligations stated up front. "Open it once a day for 14 days, tell me when something breaks" — the honest ask gets honest testers.
  • Developer communities (subreddits, Discord servers, Telegram groups around Android development). Reciprocal testing works but consumes your own hours — remember that time is a cost too.
  • A managed service. 15 real testers on real devices with daily usage logs, from $47 (about $3.13 per tester) — the same people who keep opening the app every day without you nudging them.

Set Expectations Before Day 1#

Every tester should know three things before they opt in: the app's one-line purpose, the 14-day commitment with a daily open, and where to report problems (one channel — a form, a group chat, an email address). A tester who knows what "participating" means on day 1 is still there on day 14. A tester who guessed stops opening the app by day 3.


Days 1–3: First-Session Checks (9–14)#

The first three days decide your engagement curve. Google's dashboards weight early activity heavily, so get testers into the app immediately.

  • Every tester opted in and installed. Chase non-installers on day 1, not day 7.
  • Cold start works without crash on low-end devices, not just your own flagship.
  • First-run flow completes: onboarding, login or guest mode, core action (create project, play level, send message) reachable in under two minutes.
  • Runtime permissions requested in context and the app behaves sensibly when a tester denies one.
  • No immediate uninstall spike. Android vitals → User acquisition retention will show same-day uninstalls; investigate the top reason (crash, confusion, size) today.
  • Daily open habit seeded. Send testers a nudge each morning with something new to try — a feature, a flow, a settings toggle. Twelve testers opening the app daily is the signal that passes; twelve installs that go cold is the signal that fails.

The Daily Nudge System (Templates Included)#

The single highest-leverage habit of the whole phase: one message per day, at roughly the same time, giving each tester a concrete reason to open the app today. "Please keep testing" is not a reason. "The new search screen is live, try searching with typos" is.

Why it works: engagement data aggregates daily opens. A tester without a task opens the app twice in week two; a tester with a task opens it daily. The nudge also generates feedback — each task doubles as a bug-hunt.

Templates you can adapt:

  • Day 1: "The beta is live — install from the link, complete onboarding, and tell me where you got stuck. Stuck is data, not failure."
  • Day 3: "Today: try the full core flow start to finish (create → edit → save). Time yourself. If anything takes more than a few taps, tell me."
  • Day 5: "New build is up (same link, update from Play). Kill the app, reopen it, and check your settings survived the update."
  • Day 7 (midpoint): "Halfway! Today's task: use the app with no internet for one minute, then tell me what happened. Screenshots welcome."
  • Day 10: "Edge cases day: emojis in every text field, airplane mode mid-action, double-tap everything fast. Break it for me."
  • Day 13: "Last stretch — tomorrow I submit for production. Anything that would embarrass us in public, tell me today."

Keep one channel for everything (a group chat is ideal): nudges, replies, screenshots in one place means nothing gets lost, and testers see each other participating — social proof keeps the daily curve alive better than any reminder you send privately.


Days 4–9: Engagement and Stability Checks (15–21)#

The middle week is where you convert "installed" into "used", and where stability metrics either build your case or betray it.

  • Crash-free sessions at 99.5% or better in Android vitals, per release. Any crash above 1% of sessions needs a fix uploaded to the track this week.
  • Zero ANRs — unresponsive-dialogue bugs are as damaging as crashes and easier to miss on fast devices.
  • App tested on at least 5 distinct device models across manufacturers and Android versions (14, 15, 16). Device-specific layout breakage hides until a stranger's phone finds it.
  • Offline and poor-network behavior graceful — airplane-mode tour of every core screen.
  • Battery, data and storage footprint sane. A tester noticing their battery draining is a churn you will never be told about.
  • Localization spot-check on every locale you claim to support, including RTL layout if you ship Arabic.
  • Tester feedback triaged daily. Keep a shared sheet: bug, severity, fixed-in-build. Testers who see their reports fixed keep testing — testers who feel ignored stop opening the app.

What Google Sees: Reading Your Track's Numbers#

Play Console shows your closed track's health in a handful of numbers. Knowing what "healthy" looks like turns the middle of the phase from anxiety into arithmetic — most of these live under the track's statistics and Android vitals pages.

MetricHealthyWorth worrying
Opt-in rate from your listAbove ~80% of invited testersBelow half — link or expectations problem
Installs vs. opt-insNearly everyone who opts in installsGap growing — Play Store page failing on their device
Daily active testers11–12 of 12, every dayDeclining day over day — engagement dying
Same-day uninstallsZero to oneA cluster — something crashed at first launch on a device type
Crash-free sessions99.5%+Below 99% — upload a fix this week, not next
ANR rateZeroAny sustained ANRs — jank reads like bugs to reviewers
Device spread5+ models across brandsEverything on one model — blind spots

Two readings matter most. First, trends beat snapshots: twelve testers at 100% daily on day 3 and dropping by day 8 is a worse story than a steady plateau — the trajectory is what the production evaluation sees. Second, cohort honesty: if two testers uninstall on the same day, do not top up the count and move on; find what happened that day (a release? a crash on their device model?) because the same pattern may be waiting in production where it costs real users.

Check the numbers once per day, at the same time as you send the nudge — five minutes, same order, every day.


Days 10–14: Polish and Evidence Checks (22–27)#

The final stretch is about evidence and presentation, because the production request will reference what you can prove.

  • Pre-launch report warnings reviewed — accessibility and deprecated-API warnings will not block you alone, but a pile of them weakens the quality story.
  • Vitals badges clean (no "poor behavior" flags) for the entire window.
  • Screenshots and store listing refreshed to match the final build, not your day-1 mockup.
  • Usage evidence exportable: screenshots of opt-in counts, active devices, daily engagement from Play Console — you may need them for the production questionnaire or an appeal.
  • Final regression pass on the exact bundle you intend to ship to production, on real devices.
  • Version code bumped and release notes written so your production upload is unambiguous.

Vitals Deep Dive: Fixing Mid-Test Without Breaking Momentum#

A crash in week one is normal. What separates passing campaigns from failing ones is how the fix ships.

Uploading a new bundle mid-test is allowed and expected. Bump the version code, upload to the same closed track, and testers update from their Play Store page within hours (closed-track builds skip the long production review queue). Three rules keep the fix from costing you the phase:

  • Announce every update in the nudge channel — "new build live, update and confirm version X.Y in settings". A tester stuck on the old build keeps generating old-crash data, and your vitals chart splits across two releases.
  • Watch the per-release split in vitals. After a fix lands, the old release's numbers stop moving; the new release's numbers become your reality. Judge the phase by the latest release, but keep the history — it is evidence that you found and fixed problems during testing, which is exactly what the production questionnaire asks.
  • Never re-test from zero. A fix upload does not reset your 14-day clock; the track's calendar keeps running. What resets momentum is a three-day silent gap while you "rewrite everything". Ship small fixes fast instead of waiting for a perfect build.

And one caution in the other direction: if the fix changes app behavior noticeably (new flows, renamed features), give testers a task for it the same day — untouched features mid-phase produce stale feedback and a final regression pass that finds nothing new.


After Day 14: The Production Request (28–30)#

  • All 14 days elapsed with steady activity — the requirement is 14 consecutive days of testing, not 14 calendar days with a dead middle.
  • Production access questionnaire answered honestly, in detail: who your testers were, what they did daily, what you fixed because of them. Consistency between your answers and your Console data is what manual reviewers check.
  • Staged rollout planned (20% → 50% → 100%) so any post-approval surprise is containable.

Inside the Production Access Questionnaire#

For personal accounts, the request asks you to describe the testing you ran. The reviewers — human and automated — are checking one thing: do your answers match the Console data? Write specifics, not adjectives:

  • Who tested: "12–15 external testers on physical devices (Samsung, Xiaomi, Pixel, Motorola), across Android 14–16" — not "friends and family tested it thoroughly".
  • What they did daily: "Each opened the app at least once daily for 14 days; core flows (create, edit, sync) exercised repeatedly" — not "they used it normally".
  • What changed: "Two crash clusters on Android 14 devices were fixed on day 5 (build 2.1) and one ANR on day 9 (build 2.2); testers confirmed the fixes" — not "all issues were resolved".

The failure pattern is answers that sound better than the data: claiming daily usage while the track shows dormant days, or claiming zero issues while vitals shows a crash spike. Honest and specific beats impressive and vague — the data is attached to your request whether you describe it or not.


Your Evidence Folder#

Create it on day 1 and drop things in as they happen; reconstructing history on day 14 is how inconsistencies creep in.

  • Screenshots of the track page: opt-in count, tester list, countries — on day 1 and day 14.
  • Daily engagement snapshots (or exports) from the track statistics — one per day takes ten seconds.
  • Vitals screenshots at day 7 and day 14, per release.
  • The nudge channel's message history — it documents daily testing activity and your responses to feedback.
  • Your changelog: version codes, dates, what each build fixed, which tester reports drove the fix.

If the production request is approved, this folder is trivia. If it is questioned, delayed, or appealed, it is the difference between answering in an afternoon and rebuilding your case from memory in a week.


A Realistic 14-Day Calendar#

DaysFocusChecklist items
SetupListing, bundle, testers1–8
1–3Opt-ins, first sessions9–14
4–9Stability, engagement15–21
10–14Polish, evidence22–27
Day 14+Production request28–30

Failure Patterns and What Your Data Is Telling You#

Almost every failed closed test announces itself in the data days before the production request fails. Read the symptom, fix the cause, keep the calendar:

Symptom in your trackLikely causeFix today
Opt-in rate stuck under half your listBroken join link, unclear instructions, or testers in non-available countriesRe-test the link logged-out; re-send with one-line instructions; fix country availability
Opt-ins but no installsStore listing failing on their device — unsupported version, region mismatchCheck device compatibility and track countries; ask one stuck tester for their Play Store screenshot
Installs that never openOnboarding wall — crash, permission confusion, or no obvious first taskRun first-session yourself on a low-end device; ship a fix; nudge with one concrete task
Daily opens declining from day 5No novelty — testers have nothing new to doStart the daily nudge system; ship one visible improvement per build
Uninstall cluster on one dayA crash or a bad update on that dateCorrelate with vitals per device model; hot-fix; personally re-invite the lost testers
Great usage but ANR warningsJank on a main thread — feels fine on flagships, ugly on cheap devicesProfile startup and heavy screens on the slowest device in the group
Everything healthy, then silence after day 10Tester fatigue — commitment was "a while", not "14 days"Day-13 message ("last stretch"); midpoint celebration at day 7 works too

The meta-lesson: the checklist is a early-warning system. Every item in stages 2–4 exists because some campaign discovered that failure mode the hard way. When a number drifts from "healthy", react within a day — the track runs for two weeks, and almost every recoverable failure is recoverable only early.


When the Checklist Is Too Much#

Honest self-assessment: if recruiting twelve engaged humans and running a two-week QA program competes with actually building your app, outsource the tester side. A managed closed testing service supplies 15 real testers on real devices with daily usage logs for $47 — roughly $3.13 per tester or $3.36 per day of the cycle — while you keep full control of the build. Google Play still makes the final decision; you have simply stopped gambling your launch date on strangers' goodwill.

Work the checklist, keep the daily curve alive, and day 14 stops being a cliff and becomes a formality.

⚡ Limited-time offer

Ready to Publish Your App?

Get 15 real Android testers for 14 days and meet the Google Play closed testing requirement — one-time payment.

$47

One-time payment · No subscription

  • Real devices with daily activity logs
  • Meets the 15-tester / 14-day requirement
  • Support until your app is published

Share this article