Are App Store Reviews Real? We Read 5 Million of Them

Are app store reviews real? Mostly yes — and that turns out to be the less interesting question. We pulled about 4.96 million of the most recent reviews across the iOS App Store and Google Play — 1.16 million on iOS, 3.8 million on Android — and the reviews themselves look genuine. What does not survive contact with the data is the star badge. The number under an app's icon is a lifetime average, and lifetime averages are engineered to look good. The reviews people are writing right now tell a harsher, more bimodal story.

The headline: the middle is empty. On iOS, 43% of recent reviews are five stars and 36% are one star — together that is 79% of all reviews sitting at the two extremes, with the entire 2–4 star range sharing the leftover 21%. The recent average is 3.15 stars, well below the 4.5+ badge most popular apps display. Android is gentler but the shape is the same: 62% five-star, 21% one-star, a 3.87 average, and only 17% of reviews in the middle.

The hollow middle

Here is the iOS distribution. It is not a bell curve, and it is not a slope — it is two towers with a valley between them.

5 stars43.3%

4 stars7.2%

3 stars7.0%

2 stars6.6%

1 star36.0%

This shape is not new or unique to apps. It is the J-shaped distribution of online reviews that researchers have documented for two decades — "positively skewed, asymmetric, bimodal," in the words of the foundational MIS Quarterly study: mostly five-star ratings, some one-star ratings, and hardly anything in between. An independent academic sample of 8,600 Google Play reviews found the same trough — 42.5% five-star, 8.8% two-star, 20.7% one-star. The "hollow middle" is the normal physics of a voluntary review.

What is not normal is how far our recent iOS sample tilts toward one star.

The badge is a lifetime average — and lifetime ratings lean positive

The star rating under an app is the all-time average, and the all-time average is flattering by construction. A near-census of the App Store — 62.6 million genuine reviews across 1.43 million apps — found the lifetime split was 65% five-star, 16% four-star, 6% three-star, 4% two-star, and just 10% one-star. That is the classic right-skewed J: four out of five ratings at four or five stars.

Our recent reviews look nothing like that lifetime picture — and the gap is the story.

Lifetime App Store norm (62.6M-review census)10%

all-time average — the badge

Our recent Google Play reviews21%

3.8M newest reviews

Our recent iOS reviews36%

1.16M newest reviews

A five-star app on the store can be earning one star out of every three from the people reviewing it this week. The badge is not lying, exactly — it is just answering a different question ("how has this app done over its whole life?") than the one you are usually asking ("is this app good now?").

iOS is harsher than Android

The two stores do not behave the same way. Recent iOS reviews are markedly more negative than recent Android reviews:

5 stars62.0%

4 stars8.6%

3 stars4.8%

2 stars3.8%

1 star20.7%

Android's recent reviews (62% five-star, average 3.87) sit close to the historical lifetime norm; iOS's (43% five-star, average 3.15) are the outlier. Why is genuinely open — candidate explanations include differences in when each store prompts for a review, how each surfaces the rating dialog, and which apps dominate the recent-review flow. We are flagging the gap, not claiming to have nailed its cause. But the direction is consistent with the broader record: negativity bias has been documented specifically in App Store review behavior, where consumers are disproportionately moved by negative reviews.

Why the middle is hollow: brag, moan, and the latest update

Two forces, both well studied, produce this shape — and neither requires the apps themselves to be polarizing.

Self-selection. People who write reviews are not a random sample of users. The ones who bother are disproportionately the delighted (who want to brag) and the furious (who want to moan); the merely-satisfied majority never opens the review dialog. The MIS Quarterly research calls this acquisition bias plus under-reporting bias, and it has a clean proof: when researchers forced a random group to rate the same randomly-assigned product, the bimodal spike collapsed into a normal curve with a much lower mean — only 3% five-star. The extremes you see are an artifact of who chooses to speak, not evidence that the product is loved-or-hated.

Recency. On top of that baseline, the latest reviews are event-driven. A study of 7 million App Store reviews found that a single bad update — a crash-inducing release, a forced UI redesign, a removed feature — triggers a sharp, temporary collapse in sentiment that recovers once the issue is fixed. The lifetime average barely twitches; the recent-review feed nosedives. When you read "this week's reviews," you are often reading the aftermath of the most recent release.

What earns a one-star review

The negative tower is not random venom. The dominant one-star themes in the literature — and in the searches people actually type — are remarkably consistent:

• Bugs and crashes after an update — the single most common trigger for a sentiment collapse.
• Forced changes: a UI redesign or a removed feature pushed to everyone at once. ("Why do apps update so often" and "why do apps log you out" are top Google autocompletes for a reason.)
• Pricing and subscriptions: a previously-free feature moved behind a paywall.
• Ads and account/login problems: the everyday friction that turns a 4-star user into a 1-star one.

Stars are a blunt instrument for any of this — the same research finds the star rating and the review text agree only moderately, and a meaningful share of ratings do not match the opinion written next to them. If you want the reasons, you have to read the text, which is exactly what a structured review dataset is for.

So — are the reviews fake?

This is the question people actually search ("are app store reviews fake," "are app store reviews real"), so it deserves a straight answer: the genuine ones are not the problem — and our data is a sign of that. Fake and incentivized reviews skew the opposite way from what we measured. The same App Store census found fraudulent reviews run about 70% five-star and 0.6% one-star — suspiciously glowing, almost no complaints. A recent feed carrying 36% one-star is the fingerprint of real, unhappy humans, not a manipulation farm.

The honest takeaway is not "reviews are fake." It is "the star badge is a lifetime average, and you are reading it as a verdict on the present." The reviews are real. The summary statistic is the misleading part.

How we built this — and how to rebuild it

Every figure here came from Crawlora's app-review dataset: the recent reviews of the top apps on both stores, deduplicated and queryable as JSON over one REST API — no scraping, proxies or anti-bot to manage. The same catalog backs our 4-million-app dataset and the daily App Store + Google Play charts.

Want the reasons behind the stars for one app instead of the whole market? The free App Review Analyzer summarizes the themes in any app's recent reviews, and how to scrape App Store reviews walks through the endpoint.

Read the reviews, not just the badge

Pull recent App Store and Google Play reviews for any app as normalized JSON — star, text, country, version and helpful-count — over one REST API. Proxies, rendering and anti-bot handled, billed pay-on-success. 2,000 free credits a month, no card.

Explore the app-review datasetSee the API docs