App rating calculator

From 3.8 stars to 4.5: a worked example

For 100 existing ratings averaging 3.8, the simple-average model needs 140 additional ratings averaging five to reach 4.5. If those new ratings average 4.8, it needs 234. The second scenario requires more ratings because its assumed average is closer to the target. These numbers do not mean people will choose those scores.

The formula and its limits

New average = (existing count × current average + added count × incoming average) ÷ (existing count + added count). To solve for the minimum added count, round up existing count × (target − current average) ÷ (incoming average − target). The incoming average must exceed a higher target. If you already meet the target, the model needs zero new ratings.

A displayed 3.8 may be rounded, and Apple shows territory-specific summaries. Google Play’s displayed rating can weight recent feedback. The model deliberately calculates a simple mean; compare it with your console before making a decision. Ratings and written reviews are different counts, so use the rating count.

Can one bad review change my rating?

In a simple mean, sample size determines the effect. Twenty ratings averaging four plus one one-star rating produce about 3.857. Two hundred ratings averaging four plus one one-star rating produce about 3.985. If the store rounds its summary, a small change may be invisible; the written complaint can still identify a problem to fix.

Check Apple’s ratings guidance and Google Play’s rating analysis documentation for provider-specific behavior.

What this calculation tells you

The arithmetic combines rating counts and average stars. Rounded public inputs produce estimates. Territory-specific summaries, store weighting and moderation can differ from this model. The hypothetical new average is a scenario, not a target you should purchase or require from testers.

Improve the product behind the rating

Group recurring complaints, fix blocked journeys and retest them. Ask real users for independent opinions through the platform’s supported process. Do not combine community beta scores with public store ratings or claim a simple spreadsheet predicts discovery.