Only 6.6% of Recruitment Agencies Are Built for AI Search — We Audited 391
Most recruitment agencies are invisible to AI search. In a first-party audit of 391 UK and US recruitment agencies, only 6.6% met the bar for answer-engine optimisation (AEO) — the on-page signals that get a website quoted by ChatGPT, Gemini and Google's AI Overviews. Fewer than half show even one signal, and not a single agency in the sample publishes an llms.txt file. As buyers and candidates start more of their research inside AI, that gap is quietly deciding who gets discovered.
Why AI search visibility now matters for recruitment
The first step of the buying journey has moved. When a hiring manager asks "who are the best fintech recruitment agencies in London?" — or a candidate asks "which agencies specialise in remote DevOps roles?" — a growing share of them now ask ChatGPT, Perplexity or Google's AI Overview instead of scrolling a results page. The AI answers by naming a handful of agencies and citing the sources it trusts. If your website isn't structured in a way those engines can read and quote, you're not on the shortlist — and you never find out you were left off.
This is a different game to traditional SEO. Google rewards pages that rank; answer engines reward pages they can confidently quote. That means clear factual statements, structured data, named expertise, and content shaped like the questions people actually ask. We wanted to know how ready the recruitment industry is for that shift — so we measured it.
What we audited
We built a first-party dataset of 391 small and mid-sized recruitment agencies (all 50 employees or fewer, roughly 50 per sector across eight verticals) in the UK and US, sourced from Apollo.io in June 2026. Each agency's website was scored against 34 marketing signals across five families: website and SEO foundations, lead capture, paid and martech, visitor identification, and AEO readiness.
AEO readiness was measured as six specific, observable signals that live in a page's HTML and structured data:
- FAQ / Q&A schema
- Article or BlogPosting schema
- A named author with credentials (E-E-A-T)
- An
llms.txtfile - Question-formatted headings
- Speakable or structured Q&A markup
Because these signals sit in static page source, the figures are reliable regardless of how each site renders.
The headline: under 7% are built for AI search
| Signal of AI-readiness | Share of 391 agencies |
|---|---|
| AEO-optimised (3 or more of 6 signals) | 6.6% |
| Strongly optimised (4 or more signals) | 1.5% |
| Any AEO presence (at least 1 signal) | 47.3% |
Publishes an llms.txt file | 0% |
| Uses HTTPS (table stakes, for context) | 99.5% |
Read it carefully, because the two numbers people conflate mean very different things. 47% of agencies have some AEO signal — usually one, often accidental. But only 6.6% clear the bar of being genuinely optimised (three or more signals), and under 2% are strongly optimised. The distribution is brutal: of 391 sites, 206 had zero AEO signals at all.
It's part of a wider marketing plateau
AI-readiness isn't a standalone failure — it's the leading edge of a broader one. Across the same 391 agencies, average marketing maturity scored 27.9 out of 100, and 99% sit at "Tier 2" or below — meaning a website, basic analytics and a contact form, and not much more. Visitor-identification tools are used by around 1% of agencies and marketing automation by roughly 7%. The industry's collective "that'll do" ceiling is a brochure site with a contact page.
For an ambitious agency, that's not bad news — it's the opportunity. The bar to stand out is far lower than it looks, because almost nobody has cleared it yet.
The checklist: what "built for AI search" actually means
You don't need a rebuild to become quotable. You need the signals answer engines look for. In priority order:
- Publish an
llms.txtfile. It's a plain-text summary of who you are, what you do and your key pages, written for AI crawlers. Zero of the agencies we audited had one — making it the single easiest way to stand out today. - Add FAQ and Article schema. Structured data tells engines exactly what your content is, so they can lift it with confidence.
- Write in question-and-answer shape. Headings phrased as the questions buyers ask, answered directly in the first sentence underneath.
- Show named expertise. Real authors with credentials signal the trust (E-E-A-T) that engines weight heavily.
- State facts plainly. Specific, sourced claims get quoted; vague marketing copy gets skipped.
Where to start
If you do one thing this month, publish an llms.txt and add FAQ schema to your highest-intent pages — the two cheapest moves with the biggest gap to your competitors. From there, treat every new page as something an AI should be able to quote.
This is the gap Redsun is built to close: every site we build ships these signals by default — schema, an llms.txt, named authorship and direct-answer structure — so it's answer-engine ready from day one rather than a retrofit. The agencies that move now own a position the other 93% haven't even noticed is available.