There's no form. That's the first thing to understand. You can't submit your business to ChatGPT, you can't claim a profile, and you can't pay for placement. When ChatGPT recommends a plumber, a clinic or an agency, it's assembling that answer from what it can read on the open web: your website, Bing's index, directories, review platforms, Reddit threads, Wikipedia and the rest of its training data.

Which means getting recommended isn't a trick. It's a checklist. You make your business easy to find, easy to read and easy to verify, and you do it before your competitors do. Here's the playbook we run for our own clients, in the order we run it.

First, understand how the answer gets made

ChatGPT doesn't keep a ranked list of businesses. When someone asks for a recommendation, it pulls evidence from everywhere it can reach and names the businesses whose picture is complete and consistent enough to stake an answer on. Incomplete or contradictory businesses don't get named, however good they are.

We wrote a full breakdown of where that evidence comes from in how ChatGPT decides which businesses to recommend. This article is the other half: the work, step by step.

Step 1: baseline where you stand today

Before changing anything, find out what AI currently says about you. Open a fresh ChatGPT chat and ask the questions your customers would ask: "best [your service] in [your town]", "who should I use for [the problem you solve]", "is [your business name] any good?"

Three outcomes. You're named and described accurately: protect and extend that. You're named but described wrongly: you have a consistency problem, see step 4. You're not named at all: everything below is your to-do list. Write the answers down. This is your before photo.

Step 2: get into Bing's index

The single highest-leverage hour of work on this list. ChatGPT's live web search runs on Bing's index. So do Copilot, DuckDuckGo and parts of Perplexity. A site that's invisible in Bing is invisible to most AI answer engines, regardless of how well it ranks on Google.

This is the most common gap we find when we audit small business sites. Nobody looks at Bing, because "nobody uses Bing". Your customers don't. The AI they ask does.

Do this: verify your site in Bing Webmaster Tools (you can import straight from Google Search Console in a few clicks), submit your sitemap, and claim your Bing Places listing while you're there. Then check back in a week to confirm your pages are actually being indexed.

Step 3: make your site machine-readable

An AI reads your site as source material. If your homepage opens with brand poetry and hides the facts, the AI moves on to a competitor whose page states them plainly.

Do this:

  • State the facts in the first paragraph. What you do, who for, where, from what price. Declarative sentences. "We do X for Y in Z", not "we're passionate about helping".
  • Add structured data. Organization, LocalBusiness with your address and area served, Service with prices, FAQPage on every page where questions fit. AI engines read JSON-LD directly. It's the difference between them extracting your facts and guessing at them.
  • Add an llms.txt file. A plain-text index at your site root that tells AI crawlers what your business is and where the important pages are. Ours is here as a working example. It takes an hour.

If AI engines are still not citing you after this layer is in place, the diagnostic is our list of the seven blockers. Vague titles and hedging copy are the usual suspects.

Step 4: make your facts match everywhere

AI assistants cross-reference. Your name, address, phone number, services and prices should read identically on your site, your schema, your Google Business Profile, Bing Places, and every directory you're listed in. A business whose details agree everywhere reads as verifiable. One whose details contradict each other reads as risky, and risky doesn't get recommended.

Do this: pick the canonical version of every fact, then sweep your listings and bring them into line. Claim the ones you've never touched. Most UK business owners have listings on platforms they've never logged into. Boring work, high leverage.

Step 5: build genuine third-party evidence

Your own website is a claim. Everything written about you elsewhere is evidence, and AI assistants weight evidence heavily: reviews, directory profiles, press mentions, industry listings, forum threads.

Do this:

  • Ask for detailed reviews. "Five stars, great work" tells a model nothing. A review that names the service and the town gives it quotable evidence. Change how you ask: "would you mind mentioning what we did and where you're based?"
  • Spread reviews across at least two platforms. Google reviews alone leave gaps on the platforms some engines read instead.
  • Get listed where your industry actually lives. The trade directories, professional bodies and local listings relevant to your vertical. Genuine ones only. A page of spam directory links helps nobody, and AI engines are trained on exactly the sources that ignore them.

No shortcuts here, and that's the point. This is the part competitors can't copy in a weekend.

Step 6: answer the questions your customers ask AI

People don't ask ChatGPT for keywords. They ask questions: "how much should a website cost?", "what's the difference between X and Y?", "is it worth paying for Z?" When an assistant answers, it cites the pages that answer the same question clearly.

Do this: write down the ten questions customers actually ask you before they buy. Publish a page or article answering each one, plainly, with the question in the heading and the answer in the first paragraph. One a month is enough. AI engines increasingly favour businesses with fresh, recent corroboration over ones whose trail went quiet. A steady drip beats a burst.

Honest expectations

This compounds over months. Bing indexing and site structure can move in weeks; the evidence trail takes a quarter or two to build weight. There's no instant switch, and anyone promising one is guessing or lying.

The compounding is also the good news. In most UK towns and niches, almost nobody has done this work yet. The first business in a category that becomes easy to verify tends to get named again and again, and each mention makes the next one more likely. Late movers have to displace an incumbent. Early movers just have to show up.

Recheck your baseline monthly, using the same questions from step 1. That's your scoreboard.

And if you'd rather hand the whole checklist to someone who does it every month, that's our SEO + GEO retainer: one bundled service, because about 80% of this work is shared with good SEO. Three tiers from £300/month, and every engagement starts with a free audit that shows you exactly which of these six steps you're missing.