Website services
Is your website invisible to AI search?
A site optimised properly for Google a few years ago is often invisible to ChatGPT, Perplexity, Copilot and Google's AI Overviews today. The content is there — but AI crawlers are blocked by a firewall setting, there is no llms.txt file, and the answers a customer would ask for are buried under sales copy.
01 / The health check
A scored report on how AI search sees your website.
Fifteen checks, each scored Pass, Partial or Fail. Most well-built sites from before 2024 fail ten or more.
What we check
- AI crawler access. Whether ChatGPT, Claude, Perplexity and the other AI crawlers are explicitly allowed in robots.txt
- Firewall and hosting blocks. Whether the firewall quietly blocks what robots.txt allows
- llms.txt. The one file written for AI engines specifically
- Bing indexing. ChatGPT and Copilot search through Bing, not Google
- Content without JavaScript. Many AI crawlers never run it, so hidden content is invisible
- Direct answers under headings. AI engines extract passage-level answers, not whole pages
- FAQs with schema. The most-cited content type in AI answers
- Business schema. Name, address, phone, hours and service area, complete and machine-readable
- Consistency everywhere. The same name, address and phone across your site, Google, Bing, Apple and the main directories
- Named people. An honest About page; who is behind the business is weighted heavily
- Cost guidance. A site with no prices can never be cited for “how much does it cost”
- Reviews with schema. Recommendations lean on review volume and recency
- Freshness. Visible updated dates and an honest sitemap; stale sites are deprioritised as sources
- Speed. AI crawlers have short timeouts and small crawl budgets
- Your AI visibility baseline. Real customer questions asked in ChatGPT, Perplexity, Gemini and Copilot, with every mention recorded
What you get
- A one-page scored report, dated, with a score out of 15
- A baseline of how often AI assistants mention your business today, measured with real customer questions
- Three recommended next steps, in order
02 / The action plan
Fix what the check finds.
The health check tells you where you stand. The action plan puts it right, in three stages — technical first, then content, then ongoing measurement.
Technical, done in days
- robots.txt and firewall rewrite, so AI search crawlers are explicitly allowed
- llms.txt written and published
- Bing Webmaster Tools and IndexNow set up, so new pages reach Bing and ChatGPT within hours
- A full schema layer: organisation, business, services, FAQs, reviews and people
- Your business details aligned across Google, Bing, Apple, Facebook and the main directories
Content, over two to four weeks
- Service pages rewritten so each heading is a customer question and the first paragraph answers it
- FAQ sections on every service and area page, with schema
- A cost guidance page, so the site can be cited for cost questions
- An About page with named people, experience and history
- Comparison pages for the choices customers actually weigh up
- A review capture flow, plus a reviews page with schema
- Accordions, tabs and sliders rendered as real text, not hidden behind scripts
Ongoing
- A freshness routine: visible updated dates, quarterly content refreshes
- A speed and hosting pass
- Monthly AI visibility monitoring, measured against the baseline from your check
03 / How the two fit together
The check first. The plan when you are ready.
The health check stands on its own — you can take the report and act on it yourself. If you go ahead with the action plan, the check is credited in full against it, and every package includes a 30-day and 90-day re-test, so there is a measured before and after.
Next step