Nigeria's five-star problem

The short version

The star rating on a Nigerian business is mostly a measure of how few people have reviewed it. In our directory, 5.0 is not the top of the scale. It is the default state of a listing nobody has reviewed yet, plus two friends.

That last number is the one worth sitting with. Fewer than one in twenty perfect scores has enough reviews behind it to survive a single bad week. The rest are a rounding artefact.

The distribution is broken at the top

A rating scale is useful when businesses spread across it. Nigeria’s does not. Rounded to whole stars, this is where every rated business in the directory sits:

Ninety-three per cent of rated Nigerian businesses score four stars or better. Under two per cent score below three. If you filtered a category page by “4 stars and up” you would remove almost nothing and keep almost everything, which is another way of saying the filter does no work.

This is not a Nigerian character flaw. It is what every review system does when reviewing is voluntary and mildly social. People who had an unremarkable experience do not write anything. People who were asked nicely at the counter do. The result skews up everywhere in the world. What is unusual here is how thin the evidence underneath each score is.

Two thirds of businesses are in a review desert

Before you can argue about whether a rating is honest, the business has to have one. Most do not have much:

Two out of every three Nigerian businesses in the directory have fewer than five reviews. Only 5.4% have crossed a hundred, which is roughly the point at which an average stops moving when one more person shows up angry.

So the common case is not “a business with a good rating” or “a business with a bad rating”. The common case is a business with a rating that has not been tested.

Why a 5.0 from two reviews is worse than a 4.2 from two hundred

A 4.2 average across 200 reviews has already absorbed the bad days. Some of those 200 people were served late, got the wrong order, or found the place closed when the sign said open, and the business still came out at 4.2. You are looking at a settled number.

A 5.0 from two reviews has absorbed nothing. It tells you two people were happy, or asked, or related to the owner. The next review can move it to 3.7. Choosing the 5.0 over the 4.2 is choosing the listing you know less about, purely because the number is bigger.

The practical rule: read the review count first and the average second. If the count is under five, treat the average as decoration and go and look for evidence elsewhere.

What to read instead

Three things survive this problem better than the average does.

1. Review count, then recency. Twenty reviews spread over two years beats forty from one week in 2023. A cluster of reviews on consecutive days is usually a campaign, not a trend.
2. What the negative reviews complain about. Every business has bad reviews once enough people show up. Read them for pattern, not for tone. “Rude staff” from one person is noise. Four people describing the same billing surprise is the business model.
3. Whether reviews mention specifics. Real reviews name things: the dish, the part that was replaced, the amount, how long it took, who served them. Reviews that say “good services” and “highly recommended” three times in a row are the cheapest thing to manufacture in this market.

What we do about it

This is the reason our category pages do not sort by star average. A directory that ranked on the raw number would put 45,860 barely-reviewed businesses above every established one in the country, and would be trivial to climb: get three friends to leave five stars and you outrank a hospital with 800 reviews.

We weight review substance, volume, consistency and recency, and we show the breakdown on the listing so you can disagree with us in public. The full method is at how we rank, including what we deliberately ignore.