Your Google Play rating isn't an average. Here is how it really moves.

Most "how many 5-star reviews do I need" calculators assume your rating is a simple average of every review since launch. For Google Play that assumption is wrong, and it makes a bad week look like a two-year problem.

Google is direct about it: the rating users see on Google Play is weighted towards more recent ratings, to reflect changes you make to your app. Play Console shows two numbers for this reason, your Google Play rating and your lifetime average rating. They drift apart after a bad release, and they drift back after a fix.

What the weighting does to the maths

Google doesn't publish how long the recent window is or how steeply older ratings fade, so model it simply. Say the recent window holds R ratings averaging A, and you add n new ratings averaging a. You reach a target T when:

(R × A + n × a) ÷ (R + n) = T
n = R × (T − A) ÷ (a − T)

An illustrative app shows 3.9★. About 2,000 ratings sit in the recent window, new ratings now average 4.6★, and it gets 300 a week. To reach 4.2★ it needs 2,000 × 0.3 ÷ 0.4 = 1,500 new ratings, about five weeks.

Run the same app through a lifetime calculator with its 50,000 total ratings and you get 37,500 ratings, roughly two and a half years. Teams who believe the second number give up on a recovery that was five weeks away.

One condition matters more than volume: a has to be higher than T. If your new ratings average 4.3★, you will never reach 4.5★, however many you collect. The product, not the prompt, has to change first.

What actually moves new ratings

Fix the thing people are complaining about

When Gyan Studios launched its first kids title in July 2020, the build crashed on the low-RAM phones a large share of its audience used, and the first reviews were one-star performance complaints. The app was rebuilt in 11 days, crash reports fell 91%, and the ratings curve reversed within ten days of the relaunch. No review campaign would have outrun the crash. (case study)

Reply, especially to reviews about bugs you've fixed

Users can update a rating after you reply. On a 440K MAU client app, reviews that got a reply were revised by +1.62★ on average, against +0.67★ for reviews left alone. It is the cheapest rating work there is.

Ask at the right moment, inside the rules

Google's In-App Review API lets you prompt after a good moment, such as a finished level or a completed task. Don't gate the prompt on whether the user seems happy, don't offer rewards, and don't ask repeatedly. Google Play policy prohibits incentivised or manipulated ratings, and the risk isn't worth it.

Keep the app stable

Crash and ANR rates above Google's bad-behaviour thresholds reduce visibility directly, and unstable apps collect angry reviews. The two problems feed each other.

What a recovery looks like

On a live mobile game, the store rating rose +0.48★ in a single 28-day window across about 10,500 rating contributors. The same period saw every core vital improve against the prior 28 days, including the memory-kill rate falling to 0.21%. Stability and rating moved together, which is the pattern you want. (case study)

To plan your own, use the free Google Play rating calculator. It models the recent window, shows the lifetime-average answer next to it, and projects the rating week by week. Treat the window size as your estimate, because Google doesn't publish it.

Questions people ask

Why is my Google Play rating different from my average rating?

The public rating is weighted towards recent ratings. Play Console's lifetime average counts every rating since launch, so the two diverge after any change in quality.

How many 5-star reviews do I need to raise my Google Play rating by 0.1?

With all-5-star new ratings, n = R × 0.1 ÷ (5 − T), where R is the number of ratings in the recent window and T your target. For 2,000 recent ratings and a 4.2 target, that is 250.

Does replying to reviews improve my rating?

It can, because users can revise a rating after a reply. In one client app, replied reviews moved +1.62 stars on average versus +0.67 for unanswered ones.