How to judge a supplier listing when nobody has reviewed it

How to judge a supplier listing when nobody has reviewed it

Here is a number we would rather not print. On 20 September 2026 this catalogue holds 87,205 published company listings and 12,182 published product listings, and twenty client reviews. Fourteen of those twenty are published. Six of the 12,700 product records carry a review at all, and not one of them carries two.

That is not a rounding problem. It means that for almost every listing you will see on this site, the column a buyer instinctively looks at first is empty, and will stay empty for a while. So the useful thing we can do is tell you exactly what the order you are looking at is made of, what it can support, what it cannot, and what to do instead. This piece is that.

What the ranking actually multiplies

The order of a ranked page is not a hand-placed list and it is not alphabetical. Every listing gets one composite score, and the formula is four things multiplied together.

The first is a smoothed rating. A raw average is worthless at low volume — one five-star review would put a brand-new firm above a firm with two hundred good ones — so the rating is pulled towards a prior average, and only pulled back out as real reviews accumulate. With no reviews at all, this term is the prior. It is the same number for every listing that has none.

The second is a volume term, which grows with the logarithm of the review count. Logarithms are used so that the two-hundredth review counts for far less than the second. With no reviews, this term is zero.

The third is completeness, and this is the one that matters today. It is computed live, for each row, from ten signals the listing either has or does not, and it enters the score as a multiplier between 0.5 and 1.0. A listing that documents everything is worth twice a listing that documents nothing.

The fourth is an earned-trust boost — verification, awards and the other checks a business has actually passed. Where a trust score exists it lifts the score by a bounded amount; where it does not, the multiplier is one.

Multiply those together with no reviews anywhere and the arithmetic collapses to something simple and worth saying out loud: today, the order of most pages on this site is a measure of how much each business has documented about itself, plus whatever has been independently checked. That is a real signal. It is not the signal most people assume they are reading.

The ten signals, and what each one is evidence of

The completeness term is not a vague quality judgement. It is ten specific things, each worth a tenth:

• a logo
• a description — the long or short one
• a tagline
• features listed
• pricing tiers
• integrations
• a demo video
• deployment options
• languages supported
• support channels

Read as a buyer rather than as a scoring system, that list is a decent proxy for one question: has this business decided what it sells, to whom, at what price, and said so in public? A firm that publishes its pricing tiers and its integrations has made commitments that can be checked and held against it. A firm with a logo and a paragraph has made none.

The trap to avoid is treating the inverse as damning. Plenty of excellent suppliers fill in nothing, because the work arrives by referral and the listing was never the point. A thin listing is evidence about the listing. It is weak evidence about the company, and we would rather you knew the difference than quietly benefited from your confusion.

The ten signals also do not weigh the same across kinds of supplier, and it is worth adjusting for that as you read. For a software product, pricing tiers, integrations and deployment options are close to the whole story: a vendor that will not say what it costs or what it connects to is asking you to enter a sales process to learn things its competitors publish. For a services firm, the same fields are far weaker — a development agency genuinely cannot publish a price list, and "integrations" means little when the answer is "whatever the client already runs". For those, the fields that carry real information are the description, the languages and the support arrangements, and everything else you need is on their own site rather than in any directory.

What the order does not mean

Four things the position on a ranked page is not, stated plainly because the convention in this industry is to leave them ambiguous.

It is not a quality judgement by us. Nobody here has used most of these products or hired most of these firms. We hold what businesses publish, what we can verify, and what clients tell us, and today the third of those is nearly empty.

It is not recency. A listing updated this morning and one untouched for two years are ordered by the same formula, and the ranking does not decay with age. Check for yourself that the business is still trading and still doing this work — the launch board carries a separate status for products that have been discontinued, and an archived listing stops being served entirely, but a quietly dormant agency looks exactly like an active one.

It is not paid. Sponsored placements exist, they are labelled, and they sit beside the organic list rather than inside it. Paying us cannot move a position in the ordering described above, and the fields that can move it are fields any business can fill in for free.

And it is not a substitute for the five checks below. The ordering narrows thousands of listings to a readable page. It cannot tell you whether a specific firm will deliver your specific project, because nothing that can be computed from a profile ever could.

Where our own catalogue actually stands

Since we are asking you to read the evidence, here is ours, counted on the same day from the same database.

Of the company records, 19,678 state the year the business was founded. 272 state how many people work there. 158 publish an average hourly rate and 124 a minimum project size. Those four fields are the ones a buyer uses to decide whether a firm is even in the right league, and the honest summary is that most listings do not answer them.

On the trust side: 444 records carry a computed trust score, and of the ones banded, 342 sit in "new" and 9 in "emerging". Nought have been given an award by us, because our award programmes have not run yet. One record carries a verified flag.

We also publish a separate, display-only completeness score, and it has been computed for 761 of 102,191 records — it runs when a record is touched, and most have not been touched since the feature shipped. It is not what orders a ranking and we are not going to quote it as though it were.

None of this is a good look, and printing it is the point. A directory that only publishes the numbers that flatter it is a directory whose numbers you cannot use.

What to check yourself, in the order that saves the most time

Use the ranking to build a shortlist, then do this. The order matters, because each step is cheaper than the one after it and each can end the process.

One: does the business exist the way the listing implies? Open its own site. Look for a registered company name and number, a registered address, and a jurisdiction. In the UK that number can be checked against Companies House in under a minute; most countries have an equivalent register. A supplier whose legal identity you cannot establish is not a supplier you can enforce a contract against.

Two: does it do this work, or does it list this work? Agency sites list every service a client might ask for. What you want is evidence of delivery in your specific category: named clients, dated case studies, a portfolio you can open, a repository you can read, a product you can trial. Absence here is the single most common reason a shortlist of five becomes a shortlist of two.

Three: who will actually do it? Ask for the names and the seniority of the people who would be assigned, and ask what happens if they leave mid-project. The answer separates a firm selling a team from a firm selling a placement.

Four: what does it cost, and what makes it cost more? A supplier that cannot give you a price range before a discovery call is telling you the price depends on what they think you can pay. That is not always sinister, but it is always worth knowing. Ask what typically causes a project of this size to go over, in their experience. Vague answers here are a reliable predictor of vague invoices later.

Five: ask for one reference in your own market — not their best client, but the closest one to your situation, and preferably one that has finished. Then actually call. A reference who answers "it went well" in eleven words has been briefed; a reference who tells you what was harder than expected is telling you what your project will feel like.

It is worth knowing what a good and a bad answer sound like at each step, because the difference is rarely a refusal — it is a redirection. Asked for a registered company number, a good supplier sends it in one line. A bad one sends a link to an "About us" page. Asked which of the five named clients on the site were delivered by the team that would work on yours, a good supplier names two and explains the other three. A bad one says all of them, which is almost never true of an agency that has been going more than a few years. Asked for a price range, a good supplier gives you one and names the two variables that move it. A bad one asks for your budget first, which is a pricing strategy rather than an estimate.

By the time you have done those five, the ranking has done its job: it narrowed the field. Everything that decides the outcome happens after it.

The questions that get a straight answer

Some questions are easy to answer evasively. These are harder:

• "What is the smallest engagement you would take, and the largest you have delivered?" — a range is a fact; "we are flexible" is not.
• "Which of your last five projects of this kind finished on the original budget?" — the number will not be five, and the interesting part is the explanation.
• "Who owns the code and the accounts on day one, and who owns them if we stop?"
• "What would you need from us for this to go badly?" — a supplier who can answer this has thought about failure. One who says "nothing, we handle everything" has either not run a difficult project or is not going to tell you about it.
• "Can you send the contract template before the call?" — the willingness is the answer.

None of these require you to be technical. They require you to write down the answers and compare them across three suppliers, which is a spreadsheet, not an expertise.

What we are doing about the empty column

Two things, and we would rather commit to them publicly than be asked.

We are building review collection where the evidence can be checked rather than merely typed — an invitation tied to an actual engagement, verified before it is published, with the business unable to see who said what until it is live. That is slower than opening a review box to the internet and it is the only version worth having. Until it produces volume, we will keep saying that the column is empty rather than dressing up a thin one.

And we are continuing to widen what a listing documents, because that is the part that is working: the completeness signals above are the fields we ask every business to fill, and every one of them is a claim a buyer can check.

One disclosure, since this piece is about reading evidence honestly: the company that publishes TrustList also publishes its own products, and they appear in this catalogue like any other. Where we place one of our own in a sponsored slot it carries the "Sponsored" label, and our own launches are never given editorial points on our launch board. Our listings are scored by exactly the formula described above, which — as this piece has just explained at length — currently rewards the business that documents itself most thoroughly. We are as aware as you are of who has an advantage in that, which is another reason to check the suppliers yourself rather than take an ordering on faith.