Score Your Recruitment Website Out of 100: the 34-Signal Self-Audit
This is the instrument we used to score 391 recruitment agency websites, published in full so you can run it on your own. Thirty-four binary signals across five groups. Tick what your site has today; the figure beside each item is the share of the 391 audited agencies that have it. An empty box next to a low number is an easy win few of your peers have made. An empty box next to a high number is something most already have.
It takes about twenty minutes, and every item can be verified from your own site, no tooling beyond a browser and a couple of free checkers.
01 · Website & SEO
Ten signals. These are the foundations, and they are the most widely held items in the study.
| Signal | Peers who have it | How to check |
|---|---|---|
| HTTPS / SSL | — | Your URL loads on https:// with a valid certificate |
| Mobile responsive | 81% (with SSL) | View source; look for a viewport meta tag |
| Meta description | 56% | View source; a non-empty meta name="description" |
| Optimised title tag | — | A title tag of roughly 10–65 characters, not the theme default |
| Open Graph / social tags | 61% | Paste your URL into any link-preview debugger. Does a card appear? |
| Canonical tag | — | View source; a link rel="canonical" |
| XML sitemap | 76% | Visit /sitemap.xml: it must return actual XML rather than a 404 or a styled page |
| robots.txt | — | Visit /robots.txt |
| Organization schema | 38% | Run your homepage through Google's Rich Results Test; look for Organization or LocalBusiness |
| Blog / insights hub | 44% | A link in your main navigation to /blog, /insights, /news or /resources |
02 · Lead capture
Five signals, and the group where most agency sites turn out to be thinner than their owners expect.
| Signal | Peers who have it | What counts |
|---|---|---|
| Any enquiry form | 65% | Any form element on the site |
| Genuine lead-capture form | 50% | Email plus name, phone or message. A job-search box or candidate login does not count |
| Newsletter signup | 20% | An email-only subscribe field or a subscribe call-to-action |
| Gated content / lead magnet | 2% | A report, guide or whitepaper behind a form, or a download CTA attached to one |
| Booking or multi-step form | — | A booking embed such as Calendly, or a multi-field qualifying form |
Be strict with the second row. Half the industry clears it, and the most common reason for failing is a site whose only form searches jobs.
03 · Demand & martech
Ten signals. Adoption falls away sharply after analytics.
| Signal | Peers who have it | How to check |
|---|---|---|
| GA4 analytics | 53% | View source; a gtag script with a G- measurement ID |
| Google Tag Manager | 23% | View source; a GTM- container ID |
| Any advertising pixel | 17% | Google Ads, LinkedIn Insight, Meta or Bing UET tags in source |
| Heatmap / CRO tool | 11% | Hotjar, Microsoft Clarity, FullStory or Mouseflow |
| Marketing automation | 7% | HubSpot, Marketo, Pardot or ActiveCampaign scripts |
| Live chat / chatbot | 3% | Drift, Intercom, Tawk.to, HubSpot chat or Zendesk |
Individual pixel signals, LinkedIn Insight (7%), Google Ads (10%), Meta (4%), Bing UET (1%), are scored separately in the full model, alongside other retargeting pixels.
04 · Identification
| Signal | Peers who have it | How to check |
|---|---|---|
| Visitor-ID tool | 1% | A Leadfeeder/Dealfront, Lead Forensics, Albacross or similar reveal script in source |
One row, and it carries 15 of the 100 points on its own. Across all 391 agencies it appeared on four sites. If you can tick this, you are in rarer company than any other item on this scorecard.
05 · AEO readiness
Six signals, worth 20 points. This is the group where the gap between "nobody has it" and "it is hard" is widest. Most of these are configuration work.
| Signal | Peers who have it | How to check |
|---|---|---|
| FAQ / Q&A schema | 3% | Rich Results Test; look for FAQPage, QAPage or Question |
| Article / BlogPosting schema | 7% | Rich Results Test on any blog post |
| Author / E-E-A-T markup | — | Person schema for the author, plus a visible byline and bio |
| llms.txt present | 0% | Visit /llms.txt. It must return a real file |
| Question-formatted headings | 41% | At least two H2 or H3 headings phrased as actual questions |
| Speakable / structured Q&A | — | Speakable schema, or genuine on-page FAQ blocks |
Not one of the 391 agencies audited published an llms.txt. It is a plain text file listing who you are and your key pages, it takes under an hour, and it is currently the only item on this scorecard where the whole industry sits at zero.
Score yourself
Count your ticks.
| Ticks | Tier | Where that puts you |
|---|---|---|
| Under 6 | Tier 0–1 · Foundations | Where most of the industry sits: 28% at Tier 0 and 54% at Tier 1 |
| 6–11 | Tier 2 · Active marketer | 17% of agencies, 66 of 391 |
| 12+ with visitor-ID, automation and AEO | Tier 3–4 | The top 1%. Three agencies out of 391 reached Tier 3; none reached Tier 4 |
The qualifier on the last row matters. Twelve ticks made up entirely of website and SEO items builds a brochure. A demand engine needs the conversion items too. What separates the top band is specifically the combination of identification, automation and AEO. Which is why almost nobody is in it.
What to fix first
Work the scorecard by scarcity. Start where the fewest peers have the signal. The items where peer adoption is lowest are the ones that differentiate you, and several of them are among the cheapest on the list.
- llms.txt (0% of peers). A text file. An hour of work. Nobody in the sample has one.
- FAQ schema (3%). If you already have FAQ content on the page, this is a markup job.
- Question-formatted headings (41%). Free. It is how you title sections you are already writing.
- Gated content (2%). The one item here that is genuine work, and the one that turns anonymous research traffic into a named contact.
- Cookie consent and security headers. Not scored on this card, but 46% of audited sites have no consent layer and 49% no security headers, fix these before you add any tracking script.
Note what is not on that list. Redesigning the site is not on it. Almost every high-value gap in this scorecard is a configuration or a content decision on the site you already have.
Where this scorecard fits
This card tells you how you compare to your peers. It does not tell you why your phone isn't ringing: for that, the seven-reason diagnostic works through traffic, landing pages, forms and tracking in order. And it deliberately ignores the candidate side of the experience, which the seven-point candidate experience audit covers.
The full scoring model, the five-tier curve and the sample construction behind every percentage above are in the benchmark of all 391 agencies.
Frequently asked questions
How do I audit my own recruitment website?
Check 34 binary signals across five groups: website and SEO, lead capture, demand and martech, visitor identification, and AEO readiness. Each is verifiable from your browser using view-source and Google's Rich Results Test.
What is a good score on a recruitment website audit?
Under 6 ticks puts you at Tier 0–1, where most of the industry sits. 6–11 ticks is Tier 2, reached by 17% of agencies. 12 or more including visitor-ID, automation and AEO is Tier 3–4, reached by 3 of 391.
What is the easiest way to get ahead of competing recruitment agencies online?
Publish an llms.txt file. None of the 391 agencies audited had one, it takes under an hour, and it tells AI crawlers who you are and which pages matter.
What counts as a genuine lead-capture form?
A form asking for an email plus at least one of name, phone or message. A job-search box or a candidate login does not count. Which is why 65% of agencies have a form but only 50% have a real one.
Do I need to redesign my website to improve my score?
Usually no. Most high-value gaps are configuration or content on your existing site: schema markup, an llms.txt, question-formatted headings, a consent layer and one gated asset.
Redsun Research, The State of Recruitment Marketing 2026: a first-party audit of 391 UK and US recruitment agencies, June 2026. Signals marked ", " are scored in the full model but their individual adoption rate was not published. Figures may be cited with attribution.