Foundations · Ratings & averages
Five-Star Rating Calculator
Enter how many reviews you have at each star level to get the average rating and the full breakdown — and, because a raw average lies when there are only a handful of reviews, a Bayesian-adjusted rating that pulls thin samples back toward the middle. So a 5.0 from three reviews stops outranking a 4.6 from three thousand.
The Bayesian rating behaves as if it starts with prior weight reviews at the prior rating, then adds yours. Bigger prior weight = more scepticism about small samples. Defaults: 10 reviews at 3.0.
Result
In plain English
A star average is a single number standing in for a whole distribution of opinions, and it can mislead in two directions: it ignores how many people rated (five stars from three buyers is barely evidence), and it flattens the shape (a calm 3.5 looks identical to half ones and half fives). Reading the breakdown, the count and an adjusted score together is far more honest than the headline figure alone.
- average (mean) rating
- The star value weighted by how many gave it: Σ(star × count) ∕ total reviews. The number you usually see, but blind to sample size.
- Bayesian-adjusted rating
- The average blended with a prior — (prior weight × prior rating + Σ star×count) ∕ (prior weight + total). It starts sceptical and is dragged toward your true average only as real reviews accumulate.
- share positive
- The percentage of reviews that are 4- or 5-star — often more informative than the mean, and the basis for a fairer way to rank items by a confidence bound.
- distribution shape
- Ratings are frequently “J-shaped” — piled at 5, with a bump at 1 — so the mean sits in a gap where almost nobody actually rated.
- ordinal, not interval
- The gap from 1 to 2 stars need not mean the same as 4 to 5, so averaging stars is a convenient fudge, not a precise measurement.
Frequently asked
How do you calculate an average star rating?
Multiply each star level by how many reviews gave it, add those up, and divide by the total number of reviews: (5×n₅ + 4×n₄ + 3×n₃ + 2×n₂ + 1×n₁) ∕ (n₅+n₄+n₃+n₂+n₁). For example 120 five-star, 45 four-star, 12 three-star, 6 two-star and 8 one-star reviews average (600+180+36+12+8) ∕ 191 = 4.38 stars.
Why does a 5.0 rating sometimes rank below a 4.7?
Because a good ranking accounts for confidence, not just the average. A perfect 5.0 from three reviews carries far less evidence than a 4.7 from three thousand, so sites use a Bayesian average (or a statistical lower bound) that shrinks small-sample scores toward a neutral value. The 5.0 only climbs the ranking once enough reviews back it up.
Is the average rating the best summary?
Often not. The mean hides how many people rated and what the spread looks like. A product split between delighted and furious customers can post the same 3.5 as one that merely satisfies everyone. Look at the full breakdown, the number of reviews, and the share of positive ratings — and remember reviews are self-selected, so the people who bother to rate are rarely a fair sample of all buyers.
What is a Bayesian average, and why use it for ratings?
A Bayesian average blends a product’s own ratings with a prior — a number of “virtual” reviews sitting at a neutral score. With only a few real reviews the prior dominates and the score stays cautious; as genuine reviews accumulate they take over and the score converges on the true average. It is how sites stop a single 5-star review from rocketing an item above well-reviewed rivals, and it is the fairer basis for ranking when review counts differ wildly.