How we score SEO agencies

Every criterion has bands and they were written down before a single agency was read. That order is the whole defence against a study that fits itself to its own answer, and it can be checked rather than taken on trust.

Proven outcomes rises with evidence quality rather than evidence volume: a named client beats an anonymous one, a stated period beats a bare percentage, and a before with an after beats a lift. This matters more here than it would elsewhere, because almost every agency in this field publishes case studies and almost all of them name the client.

AI search capability is worth 25 per cent, second only to proven outcomes, and the score comes from bestgeoagencies.co.uk, a separate study that reads each agency against seven criteria of its own and a tracked prompt set. It is a deeper piece of work on that question than a side-criterion here could be.

That study is published by Buried, which also appears in this ranking, and it rates Buried highest. We are not going to pretend that is a neutral arrangement. It is stated on the ranking itself, beside the column, and the underlying study publishes its own working.

Eleven of the twenty appear in it and carry their score across. The nine that do not are assessed here against the same four measures it weighs, and every row says which of the two produced its number. The second route cannot run the tracked prompt set, which is the one thing it does not reproduce.

Reviews are read from each agency’s own Google profile and nowhere else. A business that gathers reviews on one platform and not another is telling you about its marketing, and choosing whichever platform flatters it is not a measurement. A listed agency may submit a different profile through Get listed; none has.

Five classes, recorded against every score. First-party, something we measured. Second-party, something the agency publishes about itself. Third-party, a record held by somebody else. Absent, looked for and not found. Not reachable, the source exists and would not open.

The last two are different and the difference matters. Absent is a finding about the agency. Not reachable is a finding about us, and it is never scored as a zero.

Every case study naming a client over a stated period can be checked against that client’s own traffic data, and the strongest claim from each agency was. The verdict is one of three, never two.

Confirmed: direction and rough magnitude agree with the claim. Not confirmed by third-party data: they do not. Not checkable: the client is anonymous, no period is given, or the metric claimed is a ranking position, an impression count or revenue, which no third-party tool can see.

Not checkable is not a criticism. It is a fact about how a case study was written, not about the work behind it, and a reader should not read it as one.

From twelve lists pulled before anyone formed an opinion: eight agency roundups, two directories, the search results for the money terms themselves, and an AI answer to the question a buyer would actually ask. 131 names went into the pool. 116 are recorded with the reason they did not make it, which is the file that answers why is my agency not on here.

The rules are testable. UK-based or a UK office serving UK clients. SEO as a core service with its own page and at least one named SEO case study, so a full-service agency carrying SEO as a bullet does not qualify. Actively trading. Enough published material to research, because an agency nobody can check cannot be scored and scoring it zero would measure its website. Not a directory, marketplace or reseller.

One source counts half. A list of more than twenty-five names, or one that ranks itself first in its own ranking, is a directory in an article’s clothes or an advertisement, and is weighted accordingly.

Where more than twenty passed, the cut went on how often a name appeared across independent sources, then on UK organic traffic. Two places we used judgement instead, and both are on the record. Two agencies made the mechanical count on the three weakest lists in the pool while drawing 61 and 42 UK organic visits a month between them. An agency search engines barely surface is a hard thing to put on a list about search, so they came out.

This site is published by Buried, an agency that appears in the study and is scored by the same protocol as the other nineteen. That is stated here rather than buried in a footer.

An agency can pay to be included, which means it is read and scored like everyone else. No agency can pay for a position and none has. Where an agency lands is determined by its score.

It reads what agencies publish and what third parties record. It does not sit in their meetings.

An agency that does excellent work and writes about none of it will score below one that does good work and documents it well. That is a real limitation and it is the honest cost of a method a reader can check.

Nobody reaches 10 on proven outcomes in this sweep. That band needs three claims confirmed against third-party data and this sweep checked one per agency.