4.7 stars is
average.
92% of the top iPhone apps have 4 stars or more. So I re-scored all 4,883 of them against their own category. Watch the stars deflate.
Fig. 02 · Deflate any app
What's it really worth?
Fig. 03 · The deflation board
Robinhood has 4.3 stars. That's bottom 7% of finance apps.
Stars in, real rating out. Each app is ranked against the other top apps in its own App Store category. 3.0 is the middle of the pack.
Fig. 04 · Most inflated categories
In Food & Drink, a 4.5 is bottom 10%.
Each bar is where the middle 80% of a category's top apps sit on the 1 to 5 scale. Most of the scale is empty. 73% of all 4,883 apps have 4.5 or more, and only 2.5% are under 3.
Fig. 05 · Mine go first
I make Revid. It's a 2.6.
My products aren't on the App Store, so I ran them on Trustpilot. Revid goes against 13 other AI video tools. Trustpilot isn't squished like the App Store, it's love or hate: lots of 5 stars, lots of 1 stars, almost nothing in between.
AI video tools on Trustpilot · ranked by real rating
Left number: TrustScore. Right: real rating among these 14 tools. Bar: share of 5★ to 1★ reviews. Trustpilot pages read Oct 6, 2026.
Fig. 06 · How the math works
No vibes. Just rank.
4,883 apps from the US App Store top free, paid and grossing charts in 25 categories, read from Apple's public API on Oct 6, 2026. Apps with fewer than 50 ratings stay out of the pile.
Small apps get pulled toward the middle of their category, as if 200 middle-of-the-pack ratings were added to every app. A 5.0 from 12 people doesn't beat a 4.8 from 2 million.
Real rating = 1 + 4 × your percentile in your category. Middle of the pack is 3.0, the top app is 5.0, the last one is 1.0. Categories under 60 apps use the whole store.
Public numbers and simple math. This says nothing about whether any review is fake. A low real rating means "rated lower than its neighbours", that's all. Searches outside the snapshot use the live App Store rating against the Oct 6, 2026 spread.