Average star rating: how it is calculated and what 4.8 from 5 reviews means

Why does your 4.8 feel weaker than a competitor's 4.6?

Your shop shows 4.8 stars. A competitor shows 4.6. Yours is higher, and yet shoppers seem to trust theirs more. The difference is usually the small grey number next to the stars: you have 5 reviews and they have 200.

An average from five people and an average from two hundred are both written as one decimal, but they carry very different amounts of information. Once you have seen how the average is built, it is easy to see why.

How is an average star rating calculated?

Add up the stars from every rating and divide by how many ratings there are. Our demo shop, Glyna, had five approved reviews when we read its public review data on 6 November 2026: four people gave 5 stars and one gave 4. That is 20 + 4 = 24 stars from 5 people, and 24 ÷ 5 = 4.8.

Google does the same for local businesses. Its Business Profile help page on review scores, read on 6 November 2026, says the score is the average of all ratings published on Google for that place, on a scale of 1 to 5, and that a new review can take up to two weeks to change it. Review apps differ mainly in which reviews they let into the sum, which we come back to below.

If you want to check a number by hand, write the star counts down from 5 to 1, multiply each by its star value, add the results and divide by the total count. A spreadsheet does it with one formula, and the result should match what your review app shows to the first decimal.

How much does one bad review move the average?

A lot, while you have only a handful. Take Glyna's 4.8 from five reviews and add a single 1-star review: 25 stars from 6 people is 4.2. The same 1-star review on top of fifty reviews averaging 4.8 leaves 4.7, and on top of two hundred it still rounds to 4.8.

This is why a new shop's rating swings from week to week and an established one barely moves. It also explains why a perfect 5.0 from three reviews persuades nobody: everyone knows the next review can knock half a star off.

How many reviews before an average means something?

There is no magic count, but there is a useful rule of thumb from basic statistics. When ratings spread by about one star, which is what a normal mix of fives, fours and the odd low score looks like, the average you see can sit up to about two divided by the square root of the count away from the one you would get with every customer's opinion.

Read the table this way. With 20 reviews, a 4.6 and a 4.4 are not really different; with 100 they start to be. So when you compare your products, ignore gaps of a tenth or two between products that have fewer than a hundred reviews each. The same goes for shoppers comparing you with a competitor, which is why showing the count matters as much as the stars. How many reviews shoppers say they want before they trust a business, and which averages sell best, are covered in our checked list of review statistics and in do reviews increase sales.

How many 5-star reviews does it take to raise your average?

More than most people guess, and more the closer you get to 5. If you have n reviews averaging a and want to reach a target t, you need n × (t − a) ÷ (5 − t) new 5-star reviews, rounded up, assuming nothing lower arrives in between.

The last row is the one to notice. Going from 4.6 to 4.8 on a hundred reviews takes another hundred perfect scores, because each new 5 only pulls the average a little further towards 5. In practice some of the new reviews will be 4s and the odd 1, so the real number is higher.

There is no shortcut that keeps the number honest. Ask every customer the same way, not only the ones you expect to be happy: Google's content policy, read on 7 November 2026, lists asking selectively for positive reviews as fake engagement. Then look at what the low reviews have in common and fix that, since the cause of the next 1-star review is usually in the last few. If the question behind this is which average to aim for, do reviews increase sales covers the Spiegel Research Center's finding that purchase likelihood peaked at averages of 4.0 to 4.7, not at 5.0.

Why do some sites rank a 4.7 above a 5.0?

Because a 5.0 from three reviews tells you less than a 4.7 from two hundred, and some sites sort with that in mind. A common way is a Bayesian average: before dividing, add a set number of imaginary reviews at a typical score, so items with few reviews are pulled towards that score and items with many barely move.

An example with 10 imaginary reviews at 4.3. The product with three 5-star reviews becomes (10 × 4.3 + 15) ÷ 13 = 4.46. The product with two hundred reviews averaging 4.7 becomes (10 × 4.3 + 940) ÷ 210 = 4.68, and now sorts first. Both numbers come from running the formula, and the order flips the way most shoppers would expect.

Use this for sorting, not for the number you print. The average shown next to the stars should be the plain one, so it matches what visitors can check against the reviews and what you put in your markup. Google, for its part, describes its Business Profile score as the plain "average of all ratings published on Google". Other sites do not all publish their method, so check their help pages before you compare their numbers with yours.

How should half stars be shown?

A 4.3 has to become a row of five star shapes somehow, and there are three common ways. Some widgets round to the nearest whole star, some round to the nearest half star, and some fill the last star partly, so 4.3 shows four full stars and a third of the fifth. Partial fill is the most faithful; whole stars are the simplest to read and to make accessible.

Whichever way you draw them, print the number too. Rounding the picture can flatter a product: 4.5 drawn as five full stars looks like a perfect score. That is more than a cosmetic problem. Sahoo, Dellarocas and Srinivasan studied two years of purchases and returns at a North American specialty retailer, in a paper published in Information Systems Research in 2018. They used a jump in how that shop displayed its averages, and found that products shown with an average higher than their true rating were returned more often.

Also show how the stars are spread. Two products can both average 3.0, one because everyone found it middling and one because half loved it and half hated it. A small bar for each star value tells the shopper which it is, and it shows that your 4.8 is four fives and one four rather than something hidden.

Which reviews should count toward the average?

Every genuine review you have published, good and bad. If you hold back the 1-star reviews and publish the rest, the average goes up and stops meaning anything. In the US it is also a legal problem: the FTC rule on reviews forbids presenting the reviews you display as most or all of them while you hold some back for their low rating or negative tone, which our article on the FTC fake reviews rule explains.

Two exclusions are reasonable as long as you say what you count. Reviews still waiting for moderation are not part of the published average yet. And reviews from somewhere else, such as a marketplace, can be shown in their own list instead of being mixed into your own rating, so the number next to your stars describes one set of reviews. Label the count with what it is, for example “approved reviews: 5”, instead of a bare “5 reviews”.

How is the average shown to Google?

Through structured data on the product page: an AggregateRating with the average as ratingValue, the count as reviewCount, and the scale as bestRating 5 and worstRating 1. The average in the markup must match the one visitors see on the page. We cover the properties in detail in how to add review schema, and the reasons stars still might not appear in review stars not showing in Google.

How does Formgong calculate and show the average?

Formgong keeps a running count of approved reviews for each star value, site-wide and for each product, and divides the total stars by the total count, as above. The result is rounded to one decimal, the ordinary way, so a true 4.95 is shown as 5.0 and a 4.94 as 4.9. Only approved reviews count; pending and hidden ones never do, and reviews imported from a marketplace such as Etsy appear in their own labelled list and stay out of the average and the count.

The widget draws whole stars, rounded to the nearest one, and prints the average with one decimal and the words “approved reviews” beside them, for example “5.0 / 5 · approved reviews: 1” on Glyna's Salad Bowl page on 6 November 2026. So a 4.5 shows five filled stars next to the number 4.5; the number is the part to trust. From five approved reviews up, the widget also shows a bar for each star value, and the bars double as filters.

On Shopify and WooCommerce, Formgong adds the AggregateRating to the product's markup once the product has at least one approved review, with the average rounded to one decimal. Glyna's Salad Bowl page carried ratingValue 5 and reviewCount 1 when we read it. On a site you build yourself, the widget does not write markup, so use our review schema generator and keep its numbers in step with the widget.

You may wonder why the examples here use one small demo shop. On 6 November 2026 most review sites on Formgong with approved reviews were our own demo and test sites, and a spread of ratings built from test data would tell you nothing, so we stuck to the five reviews Glyna got from orders placed in the demo shop.

What does your rating show?

If your average does something this article does not explain, or you want a way of drawing stars we do not offer, write to us at support@formgong.com. We read every message.