The clients agencies showcase as their best work score 40/100 on AI visibility

When a marketing or SEO agency wants to win your business, it shows you its portfolio: the client it ranked #1, the case study with the traffic chart, the logo wall of businesses it's proud to have worked on. That work is real. But it was optimized for a world where customers found businesses through Google's ten blue links — not one where a growing share of them ask ChatGPT, Perplexity, Gemini or Claude "who's the best [X] near me?" and act on a single spoken-style answer.

So we ran a test. We collected 53 businesses that US and UK agencies feature as showcase clients on their own websites (28 US, 25 UK), and audited each one for AI visibility — not the agency, the client. Here is what we found, recounted from the production audit table on October 4, 2026.

The finding: showcase clients average 40/100

Read that median again: 36 out of 100. Half of the clients an agency picked as its best examples scored 36 or less. On average, when an AI engine was asked a real question a customer in that business's category and city would ask, the client appeared in fewer than 4 of 6 answers — and in 40% of cases, 2 or fewer. The businesses these agencies want the world to know they helped are, more often than not, not the ones AI names.

How to read an AI-visibility score

The score is built from appearances. Each mini-audit poses 6 AI answers to category-and-city questions; a business named in none of them scores near 0, and one named in all 6 scores between 67 and 88 depending on how prominently it is named and how many competitors share the answer. The 53 showcase clients fell into these bands:

A useful reading for an agency: a client under 20 is being named in at most 2 of 6 answers, which means the engines name a competitor instead in most of them. A client at 60 or above is a fixture. 20 of the 53 showcase clients sat in the first group; 15 sat in the second.

Why this happens even to well-marketed businesses

This isn't a knock on the agencies — it's a gap in what agency work has historically measured. AI engines don't rank a list; they synthesize an answer from what they can find and trust about a category. A business can have a fast site, clean schema, strong reviews and page-one Google rankings and still be absent from that synthesis, because the things that plausibly bear on an AI answer — consistent third-party mentions, clearly stated facts, presence in the sources engines actually cite — were never an explicit target of most SEO engagements. The result is exactly what the data shows: portfolio-grade clients with sub-40 AI visibility.

It's also invisible from the inside. AI answers change run to run (we've measured that separately), so a single check on a good day tells an owner nothing. And the engines diverge — a business can be a fixture on one engine and a coin flip on another (the engine-by-engine gap). Which is why the number is a surprise to the agency as often as to the client.

What this means if you run an agency

Some of your existing clients are likely to have this gap, and it is a concrete way to open a new conversation. You don't have to describe a problem in the abstract; you can show them, by name, which competitors the sampled AI answers recommended and whether they were named at all. Whether that becomes GEO (generative engine optimization) work is between you and the client — the likely actions (mentions, clearly stated facts, cited sources) overlap heavily with what you already do, and the report does not promise that any of them will change an answer.

That's why we built a white-label version. AskedAbout's Agency 5-pack is $249 for five full audits you can put your own logo on — each one requests up to 300 API answers (up to 25 customer questions × 4 engines × 3 samples, with completed and failed counts on the report) and shows a client their sampled mention rate per engine, the competitors named instead, and a prioritized fix plan. Run one on your best client and hand them the report.

Want to see the gap for a specific business first? Run a free check — it takes about a minute and no card.

Methodology

Between 12 and 14 July 2026 we identified 53 local businesses (28 US, 25 UK) that marketing, SEO and GEO agencies feature as case studies or named clients on their own public portfolio pages. For each business we ran an AskedAbout mini-audit: our engine poses 6 AI answers to category-and-location questions a prospective customer would ask, and scores 0–100 based on how often and how prominently the business is named versus competitors. Scores and appearance counts are the engine's own output. 12 of the 53 were run twice on the same day; the later run is used, and every one of the 53 completed all 6 answers. Recount, October 4, 2026: the July version of this page reported 45 businesses, average 40.2, median 33, 17 under 20 and 18 in 2 or fewer answers, after removing businesses that shared an agency; that agency mapping was not kept, so the figures above count every distinct business in the production audit table for those three days. The average (40.3), the under-20 share (38%), the 2-or-fewer share (40%) and the range (8 to 88) are unchanged to the rounded figure; the median moved from 33 to 36. The mini-audit is a lighter probe than our full audit (up to 300 answers) and is intended to surface the visibility gap, not to be a definitive per-engine breakdown. No client, agency or competitor is named on this page.