Look at the count, the distribution and the dates rather than the average. A rating from few reviews is weak evidence, and a perfect one from many is unusual enough to be worth examining.

Why the average is weak

It compresses everything into one figure and discards the information that would let somebody interpret it.

Four point seven could be twelve reviews or three hundred, all recent or all from 2016, evenly distributed or a mass of fives with two ones.

Those are different businesses and the average cannot distinguish them.

Which is why anybody assessing a business carefully looks past it, and why a business managing its own reputation should too.

Count as evidence

The first thing to read after the average.

Twelve reviews is a sample small enough that two unusual experiences move the figure noticeably.

Three hundred is stable, and a rating drawn from it is a genuine description of what happens rather than a snapshot of a dozen occasions.

For a small business, thirty is roughly where a rating starts being persuasive, and a hundred is where it becomes difficult for a competitor to argue with.

Which means the count is worth more attention than the decimal place, both when reading and when building.

The distribution

Where the useful information actually lives.

A four point five made of mostly fours reads differently from one made of fives and ones.

The first is a business that is consistently good and not exceptional. The second is one that is excellent most of the time and occasionally goes badly wrong.

Those need different responses from the business, and they mean different things to a customer.

Most platforms show the breakdown, and it takes ten seconds to read and tells you more than the headline figure ever will.

A worked example

A business at four point six that considered its reputation settled.

Reading the distribution properly showed a clear pattern: a mass of five-star reviews, and every single one-star review mentioning scheduling.

The average had been hiding a specific and fixable problem behind an acceptable number.

They changed how appointments were confirmed, adding a reminder the day before and a call if anybody was running late.

Over the following year the one-star reviews stopped, and the average moved to four point eight.

The useful part was not the improved figure. It was that the reviews had been telling them exactly what to fix for two years, in a form the average concealed.

Reading your own negatives as data

The most valuable use of a review page and the least comfortable.

Read every negative review you have, in one sitting, and note what each one is actually about.

Frequently three or four separate complaints turn out to be the same underlying issue: communication, timekeeping, tidiness, or a specific stage of the process.

That is free, specific, unsolicited feedback about the weakest part of the business, from people with no reason to soften it.

Most businesses read negative reviews individually, defensively, and never in aggregate, which is precisely where the pattern lives.

A perfect rating

Less persuasive than businesses assume.

A five point zero from a large number is unusual, and a proportion of readers treat it with suspicion rather than admiration.

A handful of fours among the fives reads as genuine, and the presence of a moderate review makes the positive ones more credible.

Which is a good reason not to worry about a four-star review, and certainly not to chase the person about it.

What is worth worrying about is a listing with only fives from very few reviews and no dates spread over time, which is the shape a purchased set has.

Ratings that are not comparable

A trap when comparing yourself against competitors.

Different platforms mean different things by the same number, because the populations and the prompting differ.

A platform that asks every customer automatically produces different ratings from one where only motivated people bother, and the second skews towards extremes.

Which means a four point three on one platform can represent a better business than a four point eight on another, and comparing across them is meaningless.

The comparison worth making is against competitors on the same platform, in the same area, which is the comparison a customer is making anyway.

Anything else is a number without a context, and businesses regularly worry about a figure that is perfectly normal for where it sits.

The counter-case

Where the average genuinely matters.

Where a platform uses it as a filter, so falling below a threshold removes you from results people actually see.

Where a directory or a referral scheme has a minimum rating for inclusion.

And in the immediate comparison itself, where somebody scanning three listings does read the number first, before any of the detail.

Which means the average matters for being chosen and the distribution matters for knowing what to fix, and both are worth attention for different reasons.

What customers do with the number

Less arithmetic than businesses imagine.

Most people do not distinguish between four point six and four point eight in any meaningful way. Both read as good.

What they do distinguish is above four against below four, and any rating against no rating at all.

Which means the anxiety about a decimal point is largely misplaced, and the effort belongs in the count and the recency instead.

The exception is a rating that drops below about four, where the perception changes sharply and a genuine problem is usually behind it.

Recovering an average

Arithmetic worth understanding before panicking.

With twelve reviews at four point eight, one one-star review takes you to about four point five.

With a hundred, the same review moves it by a few hundredths.

Which means the recovery from a bad review is not a matter of getting it removed. It is a matter of having enough others that it cannot move the figure.

And that work has to be done in advance, because it takes months and the bad review arrives without notice.

What to look at

  1. The count, before the average.
  2. The distribution, which shows the shape.
  3. The dates, for recency and clustering.
  4. Every negative review, read together.
  5. What the negatives have in common.
  6. Whether that thing has been fixed.

The fifth and sixth are the only parts of this list that change the business rather than simply describing it.

How to build the count that absorbs an outlier is covered in the business with no reviews at all.


Frequently asked questions

Why is the average weak?

It compresses everything into one figure. Four point seven could be twelve reviews or three hundred, recent or from 2016, evenly spread or fives and ones.

How many reviews make a rating persuasive?

Around thirty for a small business, and a hundred makes it difficult for a competitor to argue with. The count matters more than the decimal place.

What does the distribution tell me?

A four point five made of mostly fours is consistently good; one made of fives and ones is excellent most of the time and occasionally badly wrong.

What is the most valuable thing on a review page?

Your own negative reviews read together in one sitting. Three or four separate complaints are usually the same underlying issue.

Is a perfect rating best?

Not necessarily. A five point zero from many is unusual enough that some readers treat it with suspicion, and a few fours make the fives credible.

How do I recover from a bad review?

Not by removal. With twelve reviews one bad one moves the average noticeably; with a hundred it moves it by hundredths. That work has to be done in advance.

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.

Sitting at four point six and treating it as settled?

Read the distribution. The one-star reviews usually have one thing in common, and it is usually fixable.

Start a Conversation