How Your App's Average Star Rating Is Actually Calculated

Two stores, two completely different formulas. Google Play shows a *recency-weighted* average — your newest ratings count for more than the ones you earned years ago — and it recalculates that number per country and per device type. Apple's App Store shows a *cumulative* average across your app's entire history, with one escape hatch: you can reset it to zero when you ship a new version. That single difference is why a rough launch haunts an iOS app far longer than an Android one, and why a burst of fresh 5-stars moves your Play number surprisingly fast.

The practical takeaway sits underneath the mechanics: your star rating behaves less like a fixed grade and more like a rolling average of how people feel about the app *right now*. Understand which signals each store weights, and you know exactly which levers actually move the number. Here's how each formula works, and why the freshness of your reviews matters more than most teams assume.

Until August 2019, Google Play used a lifetime cumulative average — every rating an app had ever received counted equally, so a five-star you earned at launch pulled exactly as hard as one from last week. At Google I/O 2019, Google announced the switch: the rating shown on the Play Store is now recalculated to give more weight to your most recent ratings, and that change rolled out that summer.

The effect is that your current quality matters more than your history. An app that shipped buggy, got hammered with one-stars, then genuinely fixed its problems will see the displayed rating climb as newer, happier ratings accumulate and the old ones fade in influence. The reverse is just as true — coast on a stellar reputation while quality slips, and recent one-stars drag the number down faster than the old formula ever allowed.

In a later change, Google also began *personalizing* the rating: users can see an average tailored to their own country and device form factor, so two people looking at the same app from different regions or on a phone versus a tablet may not see identical numbers. There is no single global figure the way there once was. What every version of the calculation shares is that bias toward recent, relevant ratings.

Apple runs the older model. The App Store summary rating is a cumulative average of the ratings your app has collected, with no publicly documented recency weighting — a pile of one-stars from a botched 2023 launch keeps dragging the average even after you've fixed everything that caused them. History is sticky on iOS in a way it no longer is on Play.

The lever Apple hands you is the reset. In App Store Connect, you can reset your app's summary rating when you release a new version, which clears the displayed average and the rating count so the new build starts fresh. It's powerful when you've turned a genuinely troubled app around — but use it deliberately, because you also throw away your accumulated rating *count*, and a five-star average built on twelve ratings reads as thinner than a 4.6 built on ten thousand. It's a reset, not a magic wand.

Apple does surface ratings per storefront territory, so the number a user sees can vary by country. But within a territory, the summary you're trying to move is essentially that running lifetime average — which is exactly why steady, recent positive volume is the only reliable way to lift it short of a version reset.

On Google Play this is literal: recency weighting is *designed* so that a recent rating pulls harder than an equally-rated one from years back. Mechanically, a wave of new 5-stars this month outweighs the same count of 5-stars from three years ago. If you want to raise a Play rating, generating fresh positive ratings is the most direct route there is — which is the whole argument for keeping your [review velocity](/blog/review-velocity) high rather than letting feedback go quiet between releases.

On Apple, fresh reviews don't get bonus weight, but they still move the needle three ways: they add volume to the cumulative average, they sit at the *top* of your listing where prospective users actually read before installing, and they're the raw material for a version reset that makes sense. Either way, your rating behaves like a flow, not a stock. A number that's holding at 4.6 is being *held* there by a steady stream of recent happy users — the moment that stream slows and the complaints keep coming, it slides.

This is also where replying quietly enters the math. A reviewer who updates a one-star to a four-star doesn't just delete a bad rating — they mint a fresh high one, landing it exactly where recency weighting rewards it on Play and where it lifts the cumulative average on Apple. Responses change ratings often enough to matter: a study of over four million reviews by Hassan et al. found users were roughly six times more likely to raise their rating after a developer replied (4.4% versus 0.7% with no reply), and McIlroy et al. (IEEE, 2017) found that when a rating changed after a response, 38.7% of those changes were increases.

That reply is 236 characters (comfortably under Google Play's hard 350-character cap), names the exact bug and the exact fix, and gives the reviewer a concrete reason to come back and re-rate. When they do, on Play that new rating counts extra; on Apple it nudges the cumulative average up and pushes the angry review off the top of your listing. The full case for why this works is in [does replying to app reviews raise your rating](/blog/does-replying-to-app-reviews-raise-your-rating), and the turnaround script for angry reviewers is in [how to respond to negative app reviews](/blog/how-to-respond-to-negative-app-reviews).

When Google announced recency-weighted ratings at I/O 2019, it reported the same year that developers who respond to reviews see an average lift of 0.7 stars. That number gets thrown around loosely, so be precise about the mechanism: replying doesn't magically edit your average. It works because a good, fast reply increases the odds an unhappy user revises their rating upward — and on a recency-weighted store, a fresh upward revision is worth more than an old rating sitting untouched.

Speed is the underrated variable. A reviewer is most likely to update while the problem (and your fix) is still front of mind; a reply that lands a week after they've moved on rarely gets read. So the operational goal is boring but hard: answer every reviewable review, in the reviewer's language, fast, across both stores. Doing that by hand across even a few apps is where teams fall behind — which is the gap [ReplyArgus](/features) is built to close. It watches your App Store and Google Play reviews in one inbox, drafts an on-brand reply for each — grounded in your past approved replies and your store listing, already fitted to each store's limits — and lets you approve in a click or opt into rule-based auto-publish for the clean cases.

So the number under your app icon isn't a fixed grade — it's a rolling read on how people feel about the app lately, weighted toward the newest voices on Play and stacked cumulatively (with a reset lever) on Apple. Either way, the lever you actually control is the same: keep fresh, positive, well-answered reviews flowing. [Start free with ReplyArgus](/signup) — no card — and Argus drafts your first reply in minutes, in the reviewer's own language, already sized for each store, so the reviews that move your rating never sit unanswered.