AI Search Case Study: A £1.4m Instruction That Started Inside ChatGPT.
One AI recommendation. One £1.4m sale. No ad spend behind the enquiry.
An estate agent I work with recently sold a £1.4 million property. He won that instruction because someone asked ChatGPT who the good agents were in his area, and his name came back.
No ad. No portal. No referral. An AI assistant recommended him to a stranger, and the stranger became a seven-figure listing.
Here’s the part most agents haven’t clocked: this is not a fluke, and it is not a new discipline. It’s the same local search work, arriving through a door nobody was watching.
The Situation.
An estate agent already investing in his online presence – and understandably sceptical about “AI search optimisation,” which by mid-2025 had become one of the more crowded corners of the marketing-buzzword market.
The scepticism was fair. Most of what’s sold as AI SEO is either recycled advice or invented technique.
But the underlying shift is real, and it’s measurable. A growing number of people no longer open Google to find a local business. They ask an assistant, in a sentence, the way they’d ask a friend: “Who’s the best estate agent near me for selling a period property?” And they get one answer with two or three names in it – not ten blue links.
If you’re not one of those names, you don’t get a lower ranking. You get no mention at all.
What We Did.
Start with how these systems actually answer a local question. When you ask an AI assistant about local businesses, it doesn’t have a private opinion about your town. It leans on the same local data infrastructure everyone else does – heavily on Google and Bing map pack results, plus business directories, review platforms and whatever public content it can read about you.
Which means AI visibility isn’t a separate channel with its own tricks. It’s downstream of local SEO fundamentals – with one meaningful adjustment most agents miss.
1. Foundational website pages.
Service pages and area pages that state plainly what he does and where he does it. AI systems are summarising machines – they need something unambiguous to summarise. A homepage of atmospheric photography and the phrase “bespoke property solutions” gives a language model nothing to work with.
2. Both map packs optimised – Google and Bing.
This is the adjustment. Every agent who does any local SEO works on Google. Almost none touch Bing, on the reasonable-sounding grounds that hardly anyone uses it.
Hardly any people use it. Several major AI assistants lean on Bing’s index and local data to answer questions about the physical world. An unclaimed, unoptimised Bing Places listing is a hole in your AI visibility that costs an afternoon to fix – and it is very quiet, because nothing about your Google reporting will ever show it to you.
3. Local citations generated.
Consistent name, address and phone number across the directories and listing sites these systems read. Citations are unglamorous, decades-old local SEO work. They are also exactly the corroboration an AI system uses to decide whether a business is real and where it operates.
4. Review volume grown.
When an assistant has to choose two or three names from a dozen candidate agents in a town, review quantity and quality are among the few genuinely comparative signals available. This is the single biggest lever on whether you make the shortlist – and it’s the one most agents leave entirely to chance.
5. Regular social content.
Ongoing, crawlable evidence of an active local expert. Content that demonstrates what he knows about his own market, published consistently, so that everything a system can read about him points the same way.
The Result.
A £1.4m property sold, from an enquiry that began with someone asking an AI assistant for a recommendation.
No attribution model would have caught that. There’s no campaign to credit, no click to trace, no line item. Had we not asked the seller how they found him, it would have been logged as “word of mouth” – and the channel that produced a seven-figure instruction would have stayed invisible.
That’s worth pausing on. Ask every new enquiry how they found you. Some of your competitors are already being recommended by these systems and have no idea, because nothing in their reporting has a column for it.
The Takeaway For Agents.
The good news is that AI search rewards work you should be doing anyway. There is no separate playbook: build the foundational pages, optimise both map packs, sort your citations, grow your reviews, publish consistently. Do those five things properly and you become the answer.
The bad news is a first-mover window that’s closing. Right now, in most UK towns, very few agents have done this work – which means the shortlist of names an assistant returns is short and easy to get onto. That will not be true for long.
The agents who win the next few years of local instructions will be the ones who were already the answer before their competitors realised the question had changed.
When someone asks an AI assistant who the good agents are in your town, is your name in the answer? Find out how we help with our AI search for estate agent services.