How AI search is changing how customers find local businesses
The Yellow Pages transition is happening again. Businesses that ignored Google became invisible, and the same shift is happening now with AI search: ranking well on Google no longer guarantees you’re found.
Old way vs new way
The old way
A list of links. Click through a few sites to work out who’s reliable.
The new way
An answer, or a short list. The less clear and complete your information is, the less the AI has to work with when comparing you against alternatives.
Google is now two systems, not one
Ranking earns a position on a page. Being recommended earns a place in the answer. You can have both, or just one.
A mention doesn’t always come with a link back. It still puts your name in front of people, even without the click.
What your website should already be doing for SEO
Real content, not filler
Written by someone who knows the business.
Clear headers
Organised so people and search engines can follow it.
Some authority
Backlinks and a consistent presence, not a brand new site.
Technically sound
Fast, and nothing broken under the bonnet.
Basic schema in place
The core details, name, address, hours, marked up so Google can read them properly.
What your website needs on top for AI search
Answers formatted for extraction
Direct answers up top, not buried in paragraphs.
Facts an AI doesn’t have to guess
More comprehensive schema than the basic Google minimum, alongside llms.txt and llms-full.txt, spelling out what you do and where.
Other people backing your story
Not just what your own website claims about itself.
Reviews: wherever your customers actually leave them, kept current.
Trade or professional associations: whatever applies to your line of work, shown properly.
Qualifications: real certifications, not just a line saying “fully qualified.”
Genuine connections: suppliers, partners, other local businesses you’re actually linked to.
Content that’s current
Keep the important facts about your business up to date.
Most sites have one or two SEO basics covered by accident. Almost none have the AI side sorted on purpose.
The technical layer behind AI readiness
Schema markup
Not one code block sitting on its own. A connected set of them.
The core record: exact business type (not just “business”), name, address, phone, hours, logo, images, all linked to your Google Business Profile and social accounts, not scattered and disconnected.
Where you actually work: your service area, not just your address, so an AI knows whether you cover the customer’s postcode.
People and history: the owner or founder, genuine background, real areas of expertise, properly tied to the business record rather than left as plain text.
What you actually sell or do: services and products set up as their own connected entities, not just mentioned in a paragraph.
Reviews, done honestly: genuine ratings only. Copying your Google star rating into your own schema breaks Google’s own rules and can backfire. A lot of automated “AI-ready” tools get this wrong.
Most automated tools stop at a name and an address. Done properly, the business, the website, the pages, and the services are all linked together as one connected picture, not scattered guesses.
The LLM files
Two files at the root of the site, doing two different jobs.
llms.txt: a short, curated summary. Who you are, where you are, what you do, and links through to the pages that actually matter.
llms-full.txt: the fuller version, with the real content from those pages included directly, cleaned up, no navigation clutter or cookie banners, just the substance.
Having someone with real experience keep this consistent with the rest of the site, the schema, your Google listing, everything, is exactly the detail an automated tool skips over.
These files give AI systems a clean, curated version of the important facts about your business.
Worth knowing: this is still a fairly new convention, not yet something every AI platform requires. OpenAI and Perplexity already publish their own, and adoption is growing. Not having one doesn’t stop you being recommended today, but as more sites provide this, it’s likely to become the norm rather than the exception, and getting it in place now is a head start, not a formality.
What happens as your business changes
Google and AI systems need the facts about your business to stay consistent across your website, schema and LLM files. If they start contradicting each other, you lose clarity and trust, which can make your business less likely to be recommended.
The substantial work is done when those files are first created. After that, keeping them accurate should normally just mean occasional tweaks when something important changes.
I can show you how to make simple updates yourself, do them for you on an ad-hoc basis, or keep them updated as part of a maintenance programme.
Not every AI works the same way
ChatGPT, Google, Perplexity and other AI systems don’t all find and use information in the same way. The important thing is making the facts about your business clear and consistent whichever one is looking.
What actually fixes it
Make the important information easy to find
Direct answers, descriptive headers, sections that stand alone.
Add the machine-readable information
Enhanced schema, llms.txt and llms-full.txt, checked against the real facts about your business.
Keep the words human
AI and Google both downgrade generic, AI-written text.
Stop paying to be ignored
Paid exposure isn’t the same as earning an organic AI recommendation.
Request a full website audit
I’ll review your site for SEO, AI-readiness, schema, content and technical issues, then tell you what I think needs attention.
Human-reviewed, not an instant AI report. Please allow up to 24 hours for a reply.