Average Star Rating Calculator — and reviews needed to move it

A star rating is a weighted mean, and the weights are the number of reviews at each score. That sounds obvious until you try to work out what it would actually take to lift a 4.29 to a 4.5, at which point most people guess and guess badly. This average star rating calculator does both halves of the job: it computes the exact average from your five counts, and it solves for the number of additional 5-star reviews required to reach any target you name.

Arb Digital works on local search and reputation for clients whose star rating is the first thing a customer sees, and the "reviews needed" number is the one that changes conversations. Telling a business owner they need forty more perfect reviews to move a decimal point is far more useful than telling them their rating is a little low.

What This Star Rating Calculator Does

Enter how many 1-star, 2-star, 3-star, 4-star and 5-star reviews you have. The tool multiplies each count by its score, adds the results, and divides by the total number of reviews. That is the arithmetic every review platform uses to produce the headline number on your listing, and running it yourself lets you verify a figure, model a scenario, or combine ratings from two platforms into one blended average.

The second calculation is the useful one. Given a target average, it solves for the smallest number of new 5-star reviews that would push you over the line, then reports it alongside your total review count and the share of your reviews sitting at each score. The distribution bars make the shape of the rating visible: two businesses can both show 4.3 stars while one has almost no negatives and the other has a large block of 5s masking a hard core of 1s.

How to Use It

1. Copy the five counts from your listing. Google, Trustpilot, Yelp and both app stores all publish the breakdown by star level, usually as a set of bars next to the average.
2. Check the average matches. If the figure this tool returns differs from the one on the platform, the platform is almost certainly weighting or filtering reviews rather than taking a plain mean — see the section on that below.
3. Set a target average. Pick the number you actually want to display, such as 4.5. It must be higher than your current average and lower than 5.00, since no finite number of 5-star reviews reaches a perfect score once a lower one exists.
4. Read the "5-star reviews needed" figure. That is the count of additional perfect reviews required, assuming no further negative reviews arrive in the meantime.
5. Use the distribution bars to see where the drag is coming from. A cluster at 3 stars is a different operational problem from a cluster at 1 star.

The Formula and How It Is Calculated

The average is a weighted mean. Writing n_k for the number of reviews at k stars:

Average = (1×n₁ + 2×n₂ + 3×n₃ + 4×n₄ + 5×n₅) ÷ (n₁ + n₂ + n₃ + n₄ + n₅)

With the default figures — 120 fives, 45 fours, 18 threes, 7 twos and 10 ones — the total score is 600 + 180 + 54 + 14 + 10 = 858 across 200 reviews, giving an average of exactly 4.29.

The reviews-needed calculation rearranges the same equation. If S is your current total score, N your current review count and T your target average, adding x five-star reviews gives a new average of (S + 5x) ÷ (N + x). Setting that equal to T and solving gives:

x = (T × N − S) ÷ (5 − T)

Continuing the example with a 4.5 target: (4.5 × 200 − 858) ÷ (5 − 4.5) = 42 ÷ 0.5 = 84. Eighty-four consecutive 5-star reviews, with no negatives in between, to move 4.29 to 4.5. That denominator is what makes ambitious targets expensive: as the target approaches 5.00, the divisor approaches zero and the required count grows without limit. This is the same weighted-mean logic our weighted average calculator applies to any set of values and weights.

Why Your Displayed Rating May Not Match This Number

If the average you calculate here is close to but not identical with the one on your listing, that is expected, and the difference is informative. Platforms do several things to the raw mean. Some weight recent reviews more heavily than old ones, so a business that has improved shows a rating above its lifetime mean. Some filter reviews they judge to be inauthentic, meaning the counts shown publicly may not equal the counts feeding the average. Some round to the nearest half star for display while storing more precision underneath.

Google's own documentation on review snippet and aggregate rating structured data defines an aggregate rating as an evaluation on a numeric scale and requires an actual rating value to be present for the snippet to display — a reminder that what appears in search results is generated from marked-up data, not recomputed by the search engine from individual reviews. If your rich result and your listing disagree, check what your structured data is actually publishing.

The Mean Hides the Shape, and the Shape Matters

Star ratings are famously J-shaped: people who are delighted and people who are furious both write reviews, while the large middle group of mildly satisfied customers usually does not. That produces a distribution with a tall bar at 5, a small bar at 1, and very little in between, which is a poor fit for a summary statistic designed for symmetric data. The NIST/SEMATECH statistics handbook's discussion of measures of location makes the general point directly: for skewed and heavy-tailed distributions the mean, median and mode diverge, and the mean is the measure most distorted by extreme values.

The practical consequence is that the share of 1-star reviews often tells you more than the average does. A 4.4 built from 88% fives and 8% ones is a business with a specific, repeatable failure affecting one customer in twelve. A 4.4 built from a broad spread across 3, 4 and 5 is a business that is merely unremarkable. Those need different fixes, and the average alone cannot distinguish them. The distribution bars above are there precisely so you do not have to rely on the single number.

Rounding: the Cliff Between 4.44 and 4.45

Most listings display one decimal place and many display half stars, which creates thresholds that matter far more than the underlying difference. A 4.449 average displays as 4.4; a 4.451 displays as 4.5. The gap in genuine customer experience between those two businesses is nothing, and the gap in perceived quality is a visible half-notch on a search results page. If your average is sitting just under a rounding boundary, the reviews-needed figure for the next tenth is often surprisingly small — sometimes two or three reviews rather than dozens — and that is the cheapest rating improvement available to you.

Run the calculator twice: once with your real target and once with a target just above the next rounding boundary. The difference between those two numbers frequently reframes what is worth doing this quarter.

Combining Ratings Across Platforms

Businesses often want one blended figure across Google, Trustpilot and an app store. You cannot average the three averages — that treats a listing with 12 reviews as equal in weight to one with 3,000. Add the counts at each star level across all platforms first, then run the weighted mean once on the combined totals. This tool does that correctly if you enter summed counts, and the difference from naively averaging the averages is often two or three tenths of a star.

One caveat: only combine platforms that use the same scale in the same way. A 1-to-5 star scale on one site and a 1-to-10 score on another are not interchangeable, and rescaling a 10-point score by halving it assumes the two scales are used identically by reviewers, which is rarely true. If you need to work with percentage-based satisfaction scores instead, our percentage calculator handles the conversion arithmetic, and the NPS calculator covers the separate promoter-and-detractor methodology used in net promoter scoring.

What to Do With the "Reviews Needed" Number

Treat it as a feasibility check rather than a plan. If the answer is 84 and you currently collect six reviews a month, the target is a two-year project at best, and reducing the flow of negative reviews will move the average faster than adding positive ones. If the answer is 9, it is a month's work through a straightforward request process at the point of a good customer interaction.

What the number should never become is a target for manufactured reviews. Beyond the obvious ethical and legal problems, review platforms filter aggressively and a sudden burst of uniform 5-star reviews is exactly the pattern their detection systems are built to catch. The arithmetic on this page assumes the reviews are real; it says nothing about how to obtain them.

Common Mistakes to Avoid

• Averaging the platform averages instead of pooling the underlying counts — this over-weights small listings and produces a figure that matches nothing.
• Setting a target of 5.00 — once a single review below 5 exists, no finite number of perfect reviews reaches an exact 5.00 average.
• Assuming no new negatives arrive — the reviews-needed figure is a static solve, and a real business receives both kinds of review while working toward the target.
• Reading the mean without the distribution — the share of 1-star reviews usually identifies the operational problem, and the average never does.
• Ignoring the rounding boundary — moving from 4.44 to 4.45 changes the displayed rating for a fraction of the effort a full tenth would take.

Related Free Tools From Arb Digital

Use the weighted average calculator for the same weighting logic on any data, the mean, median and mode calculator when the average alone is misleading, the NPS calculator for net promoter methodology, the churn rate calculator to connect satisfaction to retention, and the engagement rate calculator for social proof beyond reviews. The full free online tools hub has the rest.

Frequently Asked Questions

This tool performs the weighted-mean arithmetic on the counts you enter. Displayed ratings on review platforms are governed by those platforms' own weighting, filtering and rounding rules, and their published figure is the one that applies.