Review Rating Confidence Calculator
Is a 4.6 rating from 12 reviews better than 4.4 from 300?
Usually not. A small sample can produce a high average by luck, so the fair comparison is the lower bound of each rating — what it is worth as evidence. On the figures here, 4.6 from twelve has a floor of 3.51, while 4.4 from three hundred has a floor of 4.22. The higher average is the weaker rating.
A five-star average from three reviews is not a five-star product. It is three reviews.
This gives each rating a floor - what it is worth as evidence - so two ratings can be compared honestly.
Built by Muhammad Shahzeb Anwar, MSc Industrial and Organizational Psychology, who also writes the paid version of this tool. This is the Wilson interval, the same method used to rank content by rating rather than by raw average.
Two ratings, compared on evidence
Put in each rating and how many reviews it came from.
Why the average on its own misleads
An average says nothing about how much evidence stands behind it. Two reviews at five stars and two hundred at 4.4 produce averages that can be compared arithmetically and should not be compared at all.
The fix is to ask a different question: given this many reviews at this average, what is the lowest true rating still consistent with what we have seen? That number is the floor, and it is what a rating is worth as evidence rather than what it claims.
With three reviews the floor sits a long way below the average, because three reviews are compatible with a mediocre product that got lucky. With three hundred, the floor sits close to the average, because three hundred reviews are not luck.
The Wilson interval, and why not the obvious formula
The textbook interval for a proportion breaks down at the edges: at a perfect rating it produces an interval of zero width, which would say a five-star average from two reviews is certainly a five-star product.
The Wilson score interval does not have that failure. It pulls the centre toward the middle of the scale in proportion to how little data there is, and it stays inside the possible range. It is the standard method for ranking items by rating rather than by raw average.
The rating is mapped onto a zero-to-one scale, put through Wilson, and mapped back to stars, so the output is in the units you started with.
Where this changes a decision
Choosing between two suppliers, two products, two contractors: the one with the higher average is very often the one with fewer reviews, and the ranking reverses when you compare floors.
It also matters when you are the one being rated. A new listing with a handful of glowing reviews is not yet evidence of anything, and treating it as proof invites the disappointment that follows.
What this will not tell you
This compares two ratings on the evidence behind them. It cannot detect fake reviews, it does not know whether the two sets of reviewers are comparable, and it has no view on whether the thing being rated is any good. A rating is what people said, not what is true.
Questions people actually ask
Is a 5-star rating from 3 reviews good?
It is three reviews. The lower bound of a five-star average from three reviews sits far below five, because three reviews are consistent with an ordinary product that got lucky. Compare the floor rather than the average.
How do you compare ratings with different numbers of reviews?
Compute a confidence interval for each and compare the lower bounds. The Wilson score interval is the usual method because it behaves sensibly on small samples and at perfect scores, where the textbook formula does not.
What is the Wilson score interval?
A confidence interval for a proportion that stays inside the possible range and pulls its centre toward the middle when there is little data. It is widely used to rank items by rating instead of by raw average.
Does this detect fake reviews?
No. It measures how much evidence a rating carries, not whether the reviews are genuine. A hundred fabricated reviews produce a confident-looking floor.
Cite this tool
Free to use and free to quote. If a figure here is useful in something you are writing, a link back is all that is asked.
Muhammad Shahzeb Anwar, MSc Industrial and Organizational Psychology. Review Rating Confidence Calculator. Shahzeb Speaks. https://www.shahzebspeaks.com/review-rating-confidence-calculator/
The other free tools
Each one answers a single question and refuses the ones it cannot. All free, none asks for an email address.
Comparing more than two at once?
The paid Review Rating Confidence Calculator handles a list of products or suppliers, ranks them on the floor rather than the average, and keeps the working so the reversal can be shown to somebody who does not believe it.
See the full tool