Google Play store listing experiments: rules and duration
Store listing experiments are Play Console's built-in A/B tests: you show up to two variants of your icon, feature graphic, screenshots or, in localized experiments, descriptions to a share of store visitors, and Google Play reports which version wins on installs, opens or pre-registrations.
Store listing experiments let you test changes to your Google Play listing on real visitors before you roll them out to everyone. Play Console splits a share of listing visitors between your current listing and up to two variants, then tells you which performed best on installs, opens or pre-registrations. They work on default and custom store listings.
Last checked on 5 November 2026 against Play Console Help. Google renamed "global experiments" to "default graphics experiments"; they work the same way.
What are the rules for Google Play store listing experiments?
The limits in Play Console Help are short:
| Rule | Value |
|---|---|
| Variants per experiment | Up to 2, tested against the current listing |
| Experiments at the same time | 1 default graphics experiment, or up to 5 localized experiments |
| Experiment name | 50 characters or fewer, visible only to you |
| Maximum run time | Stops automatically after 6 months |
| Listings you can test | Default store listing and custom store listings |
| Target metrics | Unique user install clicks, unique user open clicks, unique user pre-registration clicks |
| Statistical settings | Audience percentage, minimum detectable effect, confidence level |
After 6 months, no new data is collected, no variant is applied and all traffic goes back to the current listing. You can still see the finished experiment and its data.
What can each experiment type test?
There are two experiment types, and the difference decides what you can test:
- Default graphics experiment: tests the icon, feature graphic and screenshots in your default listing language. If you have localized graphics in a language, users viewing your listing in that language are left out of the experiment, even if you only test the icon.
- Localized experiment: tests the icon, feature graphic, screenshots and/or your app's descriptions in up to five languages. Only users viewing your listing in the chosen languages see the variants.
Text is only testable in a localized experiment. Play Console Help does not list the app name as a testable asset. If your listing has one language, a localized experiment runs for users who view it in that default language.
What do the target metrics and statistics settings mean?
Play Console Help defines the three target metrics:
| Metric | Definition (Play Console Help) |
|---|---|
| Unique user install clicks | Users who installed your app and did not have it installed on any other device at the time |
| Unique user open clicks | Installed users who opened your app |
| Unique user pre-registration clicks | Users who pre-registered for your app |
Older articles on the web talk about "first-time installers" and "retained first-time installers"; the current help page uses the three metrics above.
The advanced settings change how strict the test is:
- Experiment audience percentage: the share of listing impressions that see a variant instead of the current listing, split equally between variants.
- Minimum detectable effect (MDE): the smallest difference between a variant and the control that counts as a win or loss. If the measured difference is smaller, the result is a draw.
- Confidence level: how often the reported confidence interval will contain the true performance. A higher level lowers the chance of a false positive.
Google says the default settings give results "in the shortest time with reasonable accuracy", and changing them can make an experiment run longer.
How do you set up a store listing experiment?
- Sign in to Play Console and go to Grow users → Store presence → Store listings.
- In the "Experiment" column for the listing you want to test, click Set up. If the option is not available, Play Console asks you to fully roll out that store listing first.
- Enter the experiment name and choose the type: Graphics experiment on default listing or Experiment on a localized listing (then pick the languages).
- Choose the target metric. The page shows an estimate of the time and the number of installs, opens or pre-registrations the test needs.
- Optionally adjust the number of variants, audience percentage, MDE and confidence level.
- Click Next, pick the asset type to test and upload or write the variants. Google recommends changing one asset at a time.
- Save, then submit the changes for review from the Publishing overview page.
How long should a store listing experiment run?
Run it until Play Console reports a result, and plan the length from your traffic using the estimate on the setup page. There is no single right number of days, and Google does not publish one. The logic behind the estimate:
- Traffic sets the pace. An experiment needs a certain number of installs (or opens) in each arm. A listing with many daily visitors reaches it in days; a small one can take weeks.
- A smaller MDE needs more data. Detecting a small improvement needs many more visitors than detecting a large one. If the estimate is too long, test a bolder change or accept a larger MDE.
- More variants split the traffic. Two variants each get half the experiment audience, so each needs longer than a single variant would.
- A higher audience percentage is faster but shows the untested variant to more people.
- Whole weeks avoid weekday bias. Weekend visitors can behave differently from weekday visitors, so run whole weeks.
- Avoid unusual periods. A holiday, a feature on Google Play or a big ad campaign changes who visits, and a result from that week may not hold afterwards.
If the result is "More data needed", come back later; Google says it will tell you if a result is expected in less than a week.
How do you read and apply the results?
Results are on the same Store listings page: the Experiments tab lists all of them, and each listing's "Experiment" column links to its own. The "Results" section explains the outcome with its confidence interval and MDE and may recommend an action:
- A variant performed well: you may see a recommendation to apply it.
- More than one variant beat the current listing, or the result is a draw: review the numbers and decide which to apply.
- The current listing performed best: click Keep current listing.
An experiment measures installs or opens from your listing. It does not measure search rank. If a winning variant changes your description, check your keyword positions in the weeks after you apply it.
How do Google Play experiments compare with Apple product page optimization?
| Google Play store listing experiments | Apple product page optimization | |
|---|---|---|
| Variants | Up to 2 | Up to 3 treatments |
| Tests at a time | 1 default graphics or up to 5 localized | 1 |
| Maximum length | Stops automatically after 6 months | Up to 90 days |
| What you can test | Icon, feature graphic, screenshots; descriptions in localized experiments | App icon, screenshots, app previews |
| Text | Descriptions, in localized experiments | Not testable |
| Result metric | Unique user installs, opens or pre-registrations | Conversion rate, improvement and confidence per treatment |
| Statistics you set | Audience percentage, MDE, confidence level | Traffic proportion; Apple recommends acting at 90% confidence or more |
| Duration estimate | Shown on the setup page | Optional estimate during setup |
| Review | Changes are submitted for review | New metadata needs App Review; alternate icons must be in the app binary |
| Change after start | Not described in Play Console Help | Not possible once the test has started |
How Uptika helps
Uptika has no A/B duration calculator; Play Console's own estimate on the setup page is the right tool for that. Before you test text, run each variant through the free Metadata Linter so a winning variant cannot carry a "best", "#1" or a repeated word into your listing. After you apply a winner, the free Keyword radar reads your Google Play rank for each keyword every day, so you can spot a search side effect that the experiment cannot show. Related guides: short description, custom store listings and measuring ASO impact. Start free.
Questions
What are Google Play store listing experiments?
They are A/B tests built into Play Console. You create up to two variants of store listing assets, Google Play shows them to a share of visitors next to your current listing, and the results show which version performs best on the metric you chose.
How many variants can a Play experiment have?
Up to two variants against the current listing. Per app you can run one default graphics experiment or up to five localized experiments at the same time.
How long should a store listing experiment run?
Until Play Console declares a result. The setup page estimates the time and the number of installs, opens or pre-registrations needed, based on your settings. Run whole weeks so weekday and weekend visitors are both included; experiments stop automatically after 6 months.
What is minimum detectable effect?
It is the smallest difference between a variant and the current listing that the experiment will treat as a real win or loss. A smaller difference counts as a draw. Setting it lower needs more visitors and more time.
Can I test my app title in a store listing experiment?
Play Console Help lists the icon, feature graphic, screenshots and, in localized experiments, your app's descriptions. It does not list the app name as a testable asset.