Recording sources in agency reviews of SEO tools

What to take away

What does this evidence assessment cover?

This article assesses the provenance of claims that appear in agency-focused SEO tool reviews, and the way review criteria are stated. It also considers whether a review separates supplier material from third-party evidence.

It does not assess the software. No accounts were opened, no campaigns were run, no dashboards were scored, and no vendor was interviewed or briefed. A performance figure in any review should be read as a vendor claim unless the reviewer names a separate source.

This matters at renewal, when a promised outcome meets an invoice. Search results for these terms are commercial, and sponsored placements sit alongside editorial work. A framework for scoring evidence consistently appears in our guide to SEO tools reviews and comparison methods, which is worth reading before you accept a headline number.

How should you treat vendor-supplied material?

Supplier pages are primary sources for what a vendor promises. Semrush's agency solution pages frame the pitch around winning pitches, proving ROI and scaling client work. That is useful information: it shows who the product is built for and what result it is sold against.

It does not show whether the result happens. Treat the page as a claim with a known author, and record the date you read it. Vendor positioning moves faster than review articles do, so an undated screenshot proves very little.

Where does independent evidence come from?

For the operational layer, published guidance beats a comparison table. The design guidance in the GOV.UK Service Manual covers how services are made usable and indexable, which is the part of SEO work a licence does not fix.

At the infrastructure end, Google's documentation on how to reduce Google crawl rate is the primary source when servers or log pipelines struggle. Name it and date it rather than paraphrasing a blog post about it.

Neither source is about SEO software, and that is the point. Both describe the conditions a tool has to work within, so record them as context rather than independent evidence of a product's performance. For user-reported experience, reviews on G2, Capterra and TrustRadius can be logged separately, but they are not controlled benchmark tests.

How should you assess a review's criteria?

Assess a review's criteria before relying on its product conclusions. The table sets out editorial judgement about review quality, not a ranking of vendors.

The table will not choose a vendor. It fixes the order in which you examine evidence, which is the step most agency reviews skip.

How should evidence be recorded?

Keep one register per tool. Each row holds the claim, its source and the date you saw it.

Record whether the source is a vendor, regulator or your own test. Add columns for the claim, source, source type, date checked, what it establishes and what it cannot establish. Where no source exists, leave the row empty; that absence is useful.

For example, a row for Semrush's agency solution pages could record claims about winning pitches, proving ROI and scaling client work. Mark the source as vendor material: it establishes how Semrush positions the offer, not whether those outcomes occur. A second row could record that Semrush's plan page says its Free plan includes 1 demo project, 10 reports per day across all vital tools, and a limited set of AI Visibility features. Label the plan page vendor material and enter the date you personally checked it; the claim does not establish performance or client outcomes.

Review the register at renewal and flag any row whose source has changed since purchase.

As a hypothetical, a team considering a £400-a-month seat bundle can check whether promised reporting saves the hours the pitch implied; this is not a quoted vendor price. Without a register, that question is settled by memory at renewal. See our costs and budget guide.

Common questions

Do agency reviews of SEO tools count as evidence?

Only partly. They are secondary sources. They become evidence when the reviewer shows which claims came from the vendor, which came from published documentation, and which are their own judgement.

Does this review include hands-on testing?

No. It assesses sources and scoring only. A hands-on claim needs a stated method behind it, and this article makes no such claim.

What should a buyer check first?

The tool's effect on crawling and site operations, because that sits outside the dashboard. Use our SEO tools selection checklist to turn that into ordered questions before a demo. Ask what happens to crawl budget when a recommendation is applied across thousands of URLs.