How to read keyword research agency reviews in the UK

Start with the evidence that could alter your choice. Follow the branch that matches the review in front of you, then stop when a claim cannot be checked from the information available. A public rating can help identify questions to ask, but the decision should rest on material you can inspect directly.

If the page offers opinion but no work

Use a public review to decide whether a provider merits a closer look, not whether its keyword work delivered results.

Identify whether you are reading customer comments, a directory profile, editorial coverage or an endorsement selected by the agency. These sources should not be read as the same kind of evidence.

Trustpilot and Google collect opinions from customers who choose to post; Clutch, DesignRush and the Recommended Agency Register are directories that may show paid or sponsored profiles. The Drum and Prolific North are trade titles with editorial coverage, while agency-selected testimonials form another category. Keep those differences in your notes before comparing providers.

• Rule of thumb: Use a public review to find candidates, not to declare a winner.

If the ranking has no visible method

Do not treat a listing's position as proof of comparative quality if the page does not explain how its ranking or rating is produced. A sponsored, featured or partner label, repeated profile wording, or an order connected to membership may indicate promotion rather than a measured assessment.

Ask every provider the same questions and record what the listing leaves unanswered. Use the selection checklist to make the evidence request consistent, then apply the same review comparison method to each candidate. If the order remains unexplained, use the page to find names, not to justify a choice.

• Rule of thumb: An unexplained position cannot stand in for evidence.

If the stars seem to rate keyword performance

Check what the reviewer actually assessed before treating a rating as proof of keyword performance. Comments may cover process documents, tools, reporting frequency, named staff, turnaround, price clarity or contract terms. Those observations can describe a service, but they do not establish whether selected terms matched demand signals.

The review format also limits what the score can establish. Clutch gathers client ratings on matters such as quality, schedule, cost and willingness to refer, alongside written feedback. Trustpilot displays an overall star average with free-text comments. Neither format records whether selected keywords performed after publication, so separate service impressions from outcome evidence.

• Rule of thumb: A service rating is not a keyword result.

If a review announces measurable gains

Ask for the reporting period, starting point and source behind any claimed gain. Without those details, treat the figure as promotional rather than verified. A useful account identifies its measurement window and explains who supplied the numbers, instead of leaving their origin unclear.

Public reviews rarely show what happened to keyword choices after content went live, because outcome information usually sits with the client rather than the reviewer. Use the review to form questions, then request direct evidence through your own test data, a reference from a comparable client or a sample deliverable.

• Rule of thumb: Without a baseline, a performance claim is hard to judge.

If platform access is presented as proof

Treat access to a named platform as context, not proof that an agency exercises good judgement. Ask whether the provider can explain why it selected one term over another and preserve that reasoning in writing. Access to keyword research platforms does not show that a provider's choices suit your requirements.

Some providers add custom scoring to commercial data, so ask which interfaces support the work if that matters to your team. Moz documents API access for developers as one example an agency might mention. Platform access alone does not establish that its keyword choices fit your categories or business requirements.

• Rule of thumb: Ask for the reasoning behind the output, not just the tool name.

If reviews have produced a shortlist

Replace published scores with evidence tested against your own brief. A written sample, a reference conversation or a paid pilot can show whether a provider can explain its decisions. Keep a record of how each candidate moves from the initial list towards a shortlist.

Check whether candidates cover your categories, explain their data sources and plan for content that becomes stale. As the field narrows, replace ratings with answers, documents and inspectable work through vendor due diligence.

Three to five agencies provide a workable comparison; a larger field can slow a decision without improving it. Read case studies for the brief, data sources and review cycle, rather than relying on a percentage lift without a baseline. Apply the same criteria to every candidate before making a final selection.

• Rule of thumb: Match the evidence request to the risk of the purchase.

If review origins or endorsements remain unclear

Check whether a quoted review remains available on its original platform, and distinguish provider-selected endorsements from ratings gathered elsewhere. The Digital Markets, Competition and Consumers Act 2024 gives the CMA powers to act against fake or incentivised reviews, including reviews commissioned by a business about itself; consult the enacted Act.

The CAP Code applies to testimonial and review claims in advertising; see ASA advice on testimonials. These rules can inform checks of review origins and advertising claims, but they do not establish whether an agency can deliver the work your team needs. They address consumer protection law, not keyword quality.

Treat Trustpilot and Google ratings as a filter for service and communication, not as evidence of keyword quality. If a case study reports a lift without a baseline, or a provider will not share a sample, request a reference or pilot instead. Apply the same evidence standard to every candidate, however polished its profile or promotional claims may appear.

• Rule of thumb: Check where a review came from, then check the work.

If the evidence still leaves you undecided

Do not try to settle the choice by gathering more public ratings. Write down the question you cannot yet answer, then ask each candidate the same thing or request the same material. That keeps the next step focused on the uncertainty that matters to your team.

A small, fixed-scope test can resolve doubts about method or fit more directly than another set of reviews. If the remaining issue concerns service details, request material that lets you inspect the provider's reasoning against your written brief rather than relying on another general endorsement.

• Rule of thumb: Resolve uncertainty with evidence candidates can answer on equal terms.