SAVI Report · UK vs US Marketing Agencies · 2026
When a business asks ChatGPT, Gemini or Google's AI to suggest a marketing agency, the AI reads agency websites and puts a handful of names forward. We checked 11,201 agencies, 4,811 in the UK and 6,390 in the US, to see how ready each one's own website is to be one of those names. 186 are fully ready. The rest are selling a service they have not finished doing for themselves, and what they are missing is mostly small.
Agencies are not hiding from AI. The crawlers get in almost everywhere. What they find is a site built to impress a human buyer: case studies, awards, a showreel, with very little labelled so a machine can tell who the agency is, what it does and why to trust it. It is the same gap agencies are paid to close for their clients.
Every agency gets a score from 0 to 100 for how ready its website is to be found and recommended by AI. To reach the top "AI-Ready" band takes 80. 186 of 11,201 (1.7%) get there. Almost everyone else is packed into the middle.
When an AI names a few agencies, it is choosing from a field where most look almost identical to it. 2,442 agencies score between 67 and 72. A few points of labelled information is enough to step out of that crowd.
The top band is a thin crust. The highest score in the report is 85.6, 1 agencies reach 85 and none reaches 90. Just under the line, 276 agencies score between 78 and 79.9. What separates the 186 from the next 276 is a couple of points, less than the 5.2 that each label on the ladder below goes with.
Lay every UK agency's score over every US agency's and the shapes match. The gap is smallest among the best agencies (2.7 points at the top tenth, 6.2 at the bottom). The UK has more agencies a long way back.
The score is built from eight areas. Agencies do well on what their platform handles for them, such as security, crawlability and page structure, and fall down on the two areas AI leans on most: labelled information, and signs of a known, trusted brand. On both, the average agency scores under 50.
The UK trails in all eight areas, by very different amounts: labelled information by 9.6 points and platform setup by 8.1, against 1.5 on easy for AI to quote.
Each agency has a Google score and an AI readiness score, and they usually rise and fall together. When they part company, Google is almost always the one ahead. 2,654 agencies are Strong or better on Google and still below Strong on AI; 185 are the reverse. Being good at Google comes first, and it is not enough on its own: 1,475 agencies score Excellent on Google and are not AI-Ready.
1,635 agencies reach the top band on Google. 160 of them are also AI-Ready. Add conversion and the field shrinks to 8. Conversion is the wall: only 21 agencies in the whole report score 80 on it, and the best conversion score is 84.0.
Group agencies by where we found them and the spread is wider than the gap between the two countries. Agencies in the partner directories of the big marketing platforms score highest in both countries, and score about the same in both. The UK's lower average comes from its long tail of locally listed agencies.
Across every agency we checked, the US leads the UK by 4.6 points. Compare only agencies listed in the same three directories in both countries, 1,924 UK and 4,498 US, and the US lead is 1.1 points, small but real. The AI-Ready share is not: 28 of 1,924 (1.5%) in the UK against 100 of 4,498 (2.2%) in the US.
2,659 UK agencies come from a local business directory rather than a marketing platform's partner list. They average 57.7, 7.2 points behind the UK's Google Partners. Most of the UK's deficit is who is in the sample, not where they are.
The 518 UK agencies that belong to a trade body such as the IPA, the DBA or BIMA average 60.0. Agencies with no membership average 60.3. Membership of a platform's partner programme goes with a higher score; membership of a professional body does not.
The 140 agencies we found in three or more directories average 70.6, and 14 of them (10.0%) are AI-Ready. The 10,133 found in only one average 62.4, with 141 AI-Ready (1.4%). Agencies that bother to be listed everywhere tend to bother with the rest.
557 agencies carry a star rating in the directories we used, and 68% of them have a perfect 5.0. Their AI readiness score barely moves with the rating. It does move with how many reviews they have: agencies with fewer than five average 68.5, those with 50 or more average 74.5. Agencies with plenty of client feedback tend to have the better-kept websites. The star itself shows no link to the score.
Agencies write the case studies, introduce the team and answer the common questions, then stop one small step short of telling AI what any of it is.
5,505 agencies have a questions-and-answers section on their site. 69% of them never added the tag that tells AI it is there. The hard part, writing the answers, is finished. The quick part is missing.
Agencies with FAQ markup score 13 points higher on average than those without, and only 15% have it. Among the top 10% of agencies, 68% do; in the bottom half, 2%.
6,478 agencies publish a founder's or team member's bio. 18% of them never label that person so AI can read who they are. The expertise is on the page and invisible to the machine.
28% of agencies have an llms.txt file, the new guide file for AI readers, and they average 10 points more than those without. It travels with the other labelling work rather than replacing it: 74% of the top 10% have one.
Clients served, years in business, results delivered: the facts that make an agency easy to recommend are almost never published in a form AI can lift. 1,514 agencies do it. For an industry that sells measurable results, this is the open goal.
Six labels do most of the work: one that says who the agency is, one for each of its people, FAQ markup, article markup on case studies, an llms.txt file, and one figure written so AI can quote it. Count how many of the six an agency has and the 11,201 sort themselves almost perfectly.
Of the 6,250 agencies with two labels or fewer, 56% of everyone we checked, not one is AI-Ready. Of the 169 with all six, 69 are (40.8%). The average agency has 2.1 of the six. The average AI-Ready agency has 5.2.
The same ladder lifts the Google score from 55.1 to 81.8 and the conversion score from 48.6 to 69.6, so the labels mark a well-kept site as much as they cause a high AI readiness score. Either way, they are the quickest test in this report.
Agencies that sell a discipline do score better at it on their own site. The margin is single digits, and the sellers still sit in the middle bands. Of the agencies that pitch AI search on their own homepage, 55 of 612 (9.0%) are AI-Ready.
1,066 agencies sell conversion rate optimisation. Their own conversion score averages 63.9, which is Moderate. 16 agencies have "AI" in their name; none of them is AI-Ready.
Nearly half of the agencies selling AI visibility have not reached Strong themselves. 260 have no llms.txt file, 329 have not tagged an FAQ for AI, and 33 block at least one AI crawler from their own site.
Put the top 10% of agencies next to the bottom half. The difference is labelled information, almost line for line. None of it depends on the quality of the work.
For each habit, the grey dot is the average score of agencies without it and the coloured dot is agencies with it. This shows what goes together, not proof that one thing causes another: an agency that labels its people tends to do the rest as well. But every helpful habit points the same way.
Two choices stand out because so few agencies make them. 99 agencies have sites that show almost nothing until the browser runs their code, which leaves AI with a near-empty page, and 387 block OpenAI's GPTBot crawler outright. Neither group has a single AI-Ready agency.
Nearly every site on two hosted platforms, an agency site builder and a DIY site builder, has an llms.txt file, a short guide to the site written for AI. WordPress sites have one 28% of the time. The scores do not follow the file: agencies on that DIY site builder average 61.7 and only one is AI-Ready, while WordPress agencies average 65.7 and make up 128 of the 186. An llms.txt file describing an unlabelled site gives AI very little to work with.
Cities sit close together, and the order says something. In the UK, Leeds, Manchester and Sheffield lead, and London, with 915 agencies, averages 60.1. In the US the Sun Belt and Mountain West lead, and the big coastal cities sit below the national average: New York at 63.3, San Francisco at 62.1, Seattle at 61.4.
The US leads in 8 of 9 agency types. The exception is digital marketing agencies. The UK's 899 creative agencies, at 54.8, are the lowest-scoring large group in the report.
Age helps a little: agencies founded between 2015 and 2019 average 69.8, against 66.3 for those founded before 2000. Bigger is not better: a one-person agency averages 67.0 and an agency of more than 200 people 64.8, a gap inside the margin of error. The platform the site is built on moves the score by 8.5 points between the best and worst groups.
The named top 10 for each country is in the full report. On this page the leaders stay anonymous: the UK's highest-scoring agency is an SEO agency in Bristol at 84.2, and the US leader an SEO agency in Paramus, New Jersey, at 85.6.
Their Google score averages 83.9, and 86% of them are Excellent on it. 30% sell AI search as a service; 90% sell SEO.
They are search and digital specialists for the most part, in cities large and small. What they share is a website that tells AI clearly who they are, what they do and why to trust them.
They score between 75 and 79.9, one good change from the top band, and most of them are missing the same two things. 67% publish no figure AI can quote and 39% have no FAQ markup for AI. Both are content changes an agency can make in an afternoon.
One band lower, 2,312 agencies sit within five points of Strong, 21% of everyone we checked. Their gaps are the same list, missing more often, and 16% still lack even the label that says who the agency is.
Suggested citation. SearchScore (2026). The AI Visibility Report: UK vs US Marketing Agencies 2026. https://searchscore.io/savi-report/uk-us-marketing-agencies-2026/
The figures, charts and data are free to reuse under the Creative Commons Attribution 4.0 licence: credit SearchScore and link to this page.
Download the data (JSON) Charts, embed codes and media pack
Press enquiries: [email protected]