50 UK agencies, none met the Strong bar

Not one of the 50 UK agencies I scored reached the Strong band. The mean was 40.2/100. That is the headline from Best in Show, my new study of UK marketing and SEO agencies against the Modern SEO + AI Optimisation Standard.

A baseline with room to move

Best in Show is live. For this August 2026 edition, I measured 50 agency sites on 19th August 2026. This is the first edition of what I intend to make a recurring index.

The distribution is blunt. 31 agencies sit in Exposed, 11 in Absent and eight in Emerging. None reached Strong or Leading. The mean score was 40.2/100.

The top 10 deserve congratulations. Opace came first at 68.1. At 64.9, two-person shop Rapid SEO London took second place, ahead of every household name in the study. Size did not settle it.

Expertise needs evidence

44 of 50 blog posts contain no credentialed expert quote. Agencies often tell clients to demonstrate expertise. Most of the sample did not give readers or machines a named expert to trust. Their advice remains good. Their evidence does not.

43 of 50 posts contain no attributed statistics. A figure without a named source gives a reader nowhere to check it. It also gives an AI system weak material to cite. Numbers need a source.

Only two of 50 agencies maintain a genuine canonical fact base about themselves. An llms.txt file appears on 23 sites. Only two of 50 block any AI crawler. Access is not the problem. The industry wants AI to read its work, then gives the machine very little worth quoting.

My yardstick is public

The judging is openly subjective in one sense. I wrote the Modern SEO + AI Optimisation Standard, so Best in Show measures agencies against my published framework. It has 10 pillars and 32 weighted checks, with sources cited for every pillar. It also doffs its cap to some of the most respected voices in SEO, AEO and GEO, including Aleyda Solis and Shaun Anderson of Hobo Web. The rules are there to challenge.

I scored the 13 checks visible on public pages. Those checks account for 47 of the 100 weight points. I converted that subtotal into a score out of 100. I did not guess at private process or off-site work. The AI crawler access check was only partly observable, so I observed it but didn't score it. It joins the scored set next quarter, when the pipeline gains a rendering check. The view is incomplete.

It is not gospel. It is one practitioner's codification of what modern SEO and AI optimisation should include. Anyone can use the interactive scorecard to assess all 32 checks, free and without signing up.

The table names the leaders

The study page names the top 10. Ranks 11 to 50 appear only as rank, city and band. That keeps the finding useful without turning the lower half into public sport. The agencies are peers, not targets.

Private scores are available. Any agency in the study can request its result. Anyone outside the 50 can get scored too. The point is to find the gaps, not to hand out tomatoes.

The next edition will show movement

This edition sets the baseline. I plan to repeat the study each quarter, applying the same method. That will make movement measurable instead of leaving improvement to a confident case study and a flattering chart.

For now, read the full agency study or score your site with the free scorecard. If your agency appears in the index, ask me for its private result. The score is already waiting.