Most people read a review section by looking at one number and then scrolling. The number is an average, and an average is the least informative thing on the page.
For the broader context behind this business guide, read How Business Directories Actually Make Money: The Five Models.
Underneath it are three dimensions that each answer a different question: how many, how recent, and what shape. Read in that order, with the written content afterwards, a review section becomes genuinely useful. Read as a single score, it mostly tells you how old the business is.
What this is, and who it is for
This is a reading skill, aimed at anyone deciding whether to buy from a business on the strength of what other people have written about it.
There is an asymmetry here that should be admitted immediately: this site publishes no reviews, solicits none, aggregates none and displays no rating of any kind. So this is a page teaching you how to read something we deliberately do not produce. That is not a dodge. Reviews are one of the few genuine sources of information about how a business behaves after you have paid it, and being able to read one well is worth having whether or not we are the ones publishing it. The reasoning behind our own abstention is at the end, and in full at why we publish no ratings at all.
Nothing here is about detecting fabricated reviews on a particular listing. There are patterns associated with manipulation, but applying them to a specific business amounts to an accusation you cannot substantiate, and we will not teach that. What follows is about extracting the information that is genuinely present.
Volume: what a lot of reviews actually tells you
Volume is the first thing you see and the most misunderstood.
A large number of reviews is primarily a measure of exposure over time: how many customers a business has had, for how long, and how systematically it asks them to write something. It correlates with age, with size, with whether the sector has a habit of reviewing, and with whether the business has a process for soliciting. It correlates only loosely with quality.
The consequence is a bias that matters enormously if you are looking for small independent suppliers. A small business will often have very few reviews, and that is not a signal. A two-person studio that does a dozen jobs a year, well, for repeat clients who never think to write anything down, will show almost nothing. A high-throughput operation in a sector where reviews are routinely requested will show hundreds. Comparing those two numbers compares their sectors and their marketing habits, not their work.
What volume does give you is a confidence interval, informally. Three reviews tell you three people's experiences, which is a small sample from a self-selecting group. Three hundred tell you that the extremes have probably been sampled, which means the absence of any serious complaint is meaningful in a way it is not at three.
So read volume as a question about the sample, not as a score. And when it is low, move straight on to the other dimensions rather than concluding anything.
Recency: the single most informative dimension, and the most ignored
If you read only one thing, read the dates.
A business is not a fixed object. It changes hands. The person who did the good work leaves. It grows past the point where the founder touches each job. It moves premises, changes suppliers, or takes on a volume of work it cannot service. Reviews from before any of those events describe a business that no longer exists, and nothing on the page marks the transition.
This creates the most important comparison in the whole exercise: a small recent set beats a large stale one, almost every time.
Ten reviews from the last six months tell you about the business you would actually be dealing with. Four hundred reviews with the most recent one from three years ago tell you about a business that used to exist, and raise a question about what happened since. Neither the total nor the average distinguishes these two cases. Only the dates do, and dates are usually the smallest type on the page.
Practical reading order for recency:
- Sort by newest if the page allows it. This is the single highest-value control in any review section.
- Look at the gap between the most recent review and today. Then look at the gaps between the recent ones. A steady trickle is a trading business. A dense cluster followed by silence is a change of some kind.
- Read the newest five before the top five. The most prominently displayed reviews are frequently the most helpful or most positive rather than the most recent.
- Weight by decay rate. How fast a business changes varies by type. A restaurant can change completely in a season. A specialist maker doing the same thing for a decade changes slowly. Adjust how far back you are willing to read accordingly.
This is the same reasoning that makes an undated badge worthless, which is covered in what a verified badge would have to prove.
Distribution: why the shape matters more than the average
An average is a lossy summary. It compresses a whole distribution into one number, and in doing so it destroys the only information you actually needed.
Consider two businesses with an identical average. The first has almost every rating clustered in the middle, with no extremes: consistent, unremarkable, predictable. The second has most ratings at the top and a visible minority at the very bottom, with nothing in between. These are entirely different propositions. The first is a business that will probably do a fine job. The second is a business that is excellent when things go well and has a failure mode. The average cannot tell them apart, and the average is what almost everyone reads.
The shape is usually available as a small bar chart next to the score. Read it before the number.
| What the shape looks like | What it usually reflects |
|---|---|
| Concentrated in the middle, few extremes | Consistency; a predictable outcome, rarely exceptional either way |
| Top-heavy with a thin, steady tail at the bottom | A generally good service with an identifiable failure mode worth reading about |
| Split between the two extremes with a hollow middle | Outcomes that vary sharply, often by job type, location or who did the work |
| Almost entirely at the top with nothing else at all | Either a genuinely narrow, well-controlled service or a selective collection process |
That last row needs care. A near-perfect distribution can absolutely be earned, particularly by a small operator doing a small number of jobs for a well-matched clientele. It can also be produced by a business that only asks satisfied customers to write something, which is common and is not misconduct. Either way, the useful move is not to draw a conclusion but to go and read the lowest-rated reviews, which brings us to content.
Content: what a specific complaint is worth versus a generic one
Once you have the shape, stop counting and start reading. Read the low ones first.
A specific negative review is the most valuable item in the whole section. It names a thing that happened: what was ordered, what went wrong, how long it took, what the business said. Specificity is expensive to fabricate and easy to verify against your own situation, and crucially it lets you ask the only question that matters: would that particular failure affect me? A complaint about a delivery delay is irrelevant if you are collecting. A complaint about poor communication over a long project is decisive if your project is long.
A generic positive review is nearly worthless. A short, enthusiastic, detail-free comment could describe any business in any sector. It is not dishonest and it is not evidence of anything. It is just what a satisfied customer writes in fifteen seconds. A hundred of them is still fifteen seconds of information.
The same asymmetry runs the other way. A generic negative review, angry and non-specific, tells you very little except that someone was upset. And a specific positive review, describing an unusual request handled well, is as informative as a specific complaint.
Read for the details a customer would have no reason to invent: lead times, the way an unexpected problem was handled, what happened when something needed returning, whether a quoted price held.
Response: what the business's replies reveal
Where a business can respond publicly, its replies are often the most informative text on the page, because they are written by the party you would actually be dealing with.
What is worth noticing:
- Whether they reply to complaints at all. Replying only to praise is a choice, and a visible one.
- Whether the reply engages with the specific issue or restates a general policy. The difference is between a business that reads and one that processes.
- Whether it moves to a private channel appropriately. Offering to sort it out directly is a good sign. Insisting on it while refusing to acknowledge anything publicly is a different one.
- Tone under pressure. A reply that argues with a customer in public tells you how a dispute with you would go, and it is one of the few glimpses of a business under stress you will get before becoming its customer.
- Whether the same complaint recurs and the replies stay the same. That is a business that knows about a problem and has decided to keep answering for it rather than fix it, which may be entirely reasonable, and is at least a known quantity.
Responses are also where the limits of the format show most clearly. You are reading a business's public conduct in a setting where it knows it is being watched.
The limits: what no review section can tell you
Some things are simply not in there, and it is worth knowing which, so you go and find them elsewhere.
- Whether the business is still trading. A review section is a historical record. The most recent review tells you when someone last bought, not whether anyone can buy today.
- Whether the reviewers are like you. Reviews aggregate people who may have wanted something entirely different from what you want.
- What did not happen. Nobody writes a review about a job that went exactly as expected and finished on time, which is most jobs.
- Anything about the people who did not review. The set is self-selected, skewed towards strong feelings in both directions.
- Solvency, insurance, licensing or capability at a scale they have not worked at. Reviews are about past experience, not about capacity or standing.
- Whether the reviews describe the current owner, staff or premises. Ownership changes are rarely visible in a review section at all.
When those are the questions, reviews are the wrong instrument. Choosing between suppliers without ratings sets out a comparison method built only from things you can check directly.
Why we publish none of this ourselves
Everything above is worth doing, and we still do not do it here.
We publish no reviews, no ratings, no stars, no scores and no rankings. We do not solicit reviews from anyone, we do not aggregate reviews from anywhere, and there is no field in the register that could hold one. An entry either meets our entry bar or it does not. That is the only judgement here, and it is a judgement about whether a business is real, independent, reachable and trading, never about whether it is good.
The reasons are structural rather than squeamish. A rating about someone else's business is a published assertion about their reputation, and we would be publishing it about small operators with no realistic way to answer back on our pages. A rating is also the single most saleable object a listing site can own: the moment a score exists, there is something to influence, something to buy and something to sell, and every incentive in the category starts pulling in one direction. Refusing to have one closes that door permanently rather than promising not to walk through it.
There is a cost to you in this, and it is a real one. We will not tell you which of two entries is better, and you will have to do the reading yourself. The full accounting of that trade-off is at why we publish no ratings at all, and the complete list of things we will not publish is at the line.
In Plain English
What it means. Read a review section in this order: how recent, what shape, what the specific complaints say, and how the business replies. The average tells you least, and the total mostly reflects the size and age of the business.
Who it helps. Anyone about to hire or buy from a business they have not used before, particularly a small one with few reviews.
When to use it. After you have decided a business is plausible, and before you commit money.
What to do next. Sort by newest. Read the five most recent, then the lowest-rated ones, and ask whether the specific failures described would affect your situation. If the business is small and has almost nothing written about it, do not treat that as a mark against it; check something else instead. How directories work covers the neighbouring reading skills.
See why we don't rate anyone
We teach reading reviews and refuse to publish any. Why we don't rate anyone explains what that costs you, why we take the cost, and what would have to change for it to stop being true.