A customer used to type "electrician near me" or "best CRM for small teams" into Google. Now, more and more of them are asking ChatGPT instead.
And the answer ChatGPT gives them isn't a list of ten links. It's a recommendation — usually two or three names, with a short reason for each. If your name isn't one of them, you don't exist for that buyer. There's no second page to scroll to. There's just the answer.
This applies whether you're an electrician in Eastleigh, a B&B in the Cotswolds, or a B2B SaaS company in San Francisco. The mechanics are identical, the principles transfer, and the buyers are doing exactly the same thing: asking AI to make the choice for them. This is the bit nobody's really talking about yet. Which is exactly why now is the moment to do something about it.
Google gives you a list. ChatGPT gives you an answer. Different game.First — what's actually changing
Customers are asking AI for recommendations
Three things have happened quietly over the last 18 months. ChatGPT crossed 800 million weekly users. Google launched AI Overview at the top of search results. Perplexity went from a curiosity to a tool people actually use to find local businesses.
The behaviour change comes from this: AI gives a complete answer in one go. No tabs to open. No reviews to compare. No directory pages to scroll past. For a busy customer with a leaking tap or a CMO trying to pick a CRM at 11pm on a Tuesday, that's enormous. They'd rather have one decent recommendation than ten options to wade through.
Try it yourself. Open ChatGPT and type something like:
"I'm in [your town] and need an electrician for a rewire. Who would you recommend?"
or
"Best B2B field service management software for a UK-based mid-market company?"
If your name comes back, brilliant. If a competitor's does and yours doesn't, that's the gap worth closing. If nobody's name comes back and ChatGPT just lists generic categories or directories — well, that tells you something different. It tells you the answer slot is sitting wide open. Get there first.
When a buyer asks an AI for a recommendation, the AI doesn't read your website the way a human does. It looks for clear, structured answers about who you are, what you do, and where (or for B2B: what category, what scale, what differentiator).
Five things that get you mentioned
This is where most agencies start using letters. GEO. AEO. LLMO. Forget the acronyms. Here's the practical list, in plain English.
Write like you're answering a question
AI doesn't read paragraphs the way humans do. It scans for direct answers to specific questions. If your service page opens with "Welcome to our website. We are a family-run business serving the local community since 1998…" — you've already lost. Same applies if it opens "Acme Software is the leading enterprise solution for digital transformation in your vertical."
Instead, lead with the answer. "A house rewire in Southampton typically costs between £3,500 and £6,000, takes 5–7 days, and includes…" Or: "Our field service management software typically costs £45 per technician per month, supports teams from 5 to 500 users, and integrates with…" Either way — that's the kind of paragraph an AI can quote. And it does.
FAQ blocks on every important page
FAQs are AI's favourite format. They're literally a question with an answer. If you've got a service page about boiler installation, a FAQ block at the bottom answering "How long does a boiler installation take?" and "Do I need a Gas Safe engineer?" gives ChatGPT exactly what it needs to quote you. If you've got a product page for a CRM, a FAQ answering "Does it integrate with HubSpot?" and "What's the user limit on the starter plan?" does the same job.
Six to ten FAQs per service page or product page. Real questions buyers actually ask. Real answers in your own words.
Schema markup that tells AI exactly who you are
Schema is invisible code in the background of your website. It tells search engines and AI: this is a local business, here's the phone number, here's the address, here's what services we offer, here are our reviews. Or for B2B: this is a software product, here's what category, here's the pricing, here are the integrations. Without it, AI has to guess. With it, AI gets a perfect summary handed to it on a plate.
You don't need to know the technical details — just make sure your website has it. LocalBusiness, Service, FAQPage, Product, SoftwareApplication, Article — there's a schema type for almost everything. If you're not sure what's on your site, ask whoever built it. If they look blank, that's your answer right there.
Get mentioned on websites AI already trusts
AI doesn't make recommendations from thin air. It builds them from sources it trusts. For local businesses: local news sites, council websites, accreditation bodies, industry directories, review platforms. For B2B: G2, Capterra, industry analyst reports, podcast appearances, guest articles in respected trade publications. The more of these mention you accurately, the more confident the AI is in recommending you.
This isn't about quantity. Five quality mentions on relevant sites beats fifty spammy directory listings every time. For a tradesperson: local press, your trade body's directory, the council's recommended trader scheme. For a B2B company: the analyst piece in your category, a thoughtful podcast appearance, a guest essay in a respected industry publication.
Reviews that say specific things
"Great service, would recommend" doesn't help an AI. "Did our rewire in Eastleigh in four days, on budget, left the place spotless" does. Or for B2B: "Cut our scheduling admin from 8 hours a week to 1 hour. Saved us hiring a coordinator." The first kind tells the AI nothing useful. The specific kind tells it exactly what you do, who you do it for, and what changes when someone hires you.
When you ask for reviews, gently nudge customers and clients towards specifics. The work you did. The outcome. What stood out. Don't script it — that's worse than vague — but give them a starting point with a few prompted questions in the review request.
Specificity is what gets you quoted. Vagueness is what gets you ignored.Third — why now
The window is wide open
Here's the thing nobody's saying out loud. Most markets have barely woken up to AI search. Most plumbers, electricians, and builders haven't heard of GEO. Most B2B SaaS companies are still spending the majority of their marketing budget on Google paid ads while their AI search visibility sits at zero. Most B&Bs are optimising their TripAdvisor profile while ChatGPT recommends the chain hotel up the road.
Which means right now, with relatively modest effort, you can be the answer in your market. Not in five years when everyone's caught up — now. The cost of being early is small. The cost of being late will be substantial, because the businesses that own the AI mentions in 2027 will have been earning that real estate for two years already.
This is one of those rare moments where the underdog gets a head start. The big national directories are slow. The agencies are still selling the old playbook. Most boards are still asking "is AI search a real thing yet?" — by the time the answer reaches consensus, the slot is taken. A small business with a properly-built website and a bit of intention can land the answer slot in their category for the next decade.
Be the answer. Not the search result.
Google gives buyers a list. AI gives them an answer. The shift is already happening — quietly, in the background of millions of conversations a day. The businesses that notice now win. The ones that notice in 2027 spend the next five years catching up.