How to Get Your Business Recommended by ChatGPT & AI Search

ChatGPT, Perplexity, Gemini and Google's AI Overviews recommend businesses they can read, verify and quote. To get named, let the AI search crawlers into your site, keep your business details identical everywhere they appear, earn mentions on the third-party sources these engines cross-check, and write pages that answer buyer questions in the first sentence. Then test it with a fixed set of prompts every month and track AI referrals in GA4.

We did not start doing this because of a trend report. Our SEO and AIEO clients kept telling us the same thing: a new customer said ChatGPT recommended them. When we looked at why, the pattern was boring in the best way. The businesses being named were the ones the engines could check from several directions at once. This is the playbook we built from that, written for owners who want to do the work themselves or know what to ask an agency for.

It sits alongside two guides we have already published. If Google's AI answers are your main concern, read how to rank in Google AI Overviews. If you need the structured data itself, our schema markup guide has the templates. This article covers the assistants, ChatGPT first, and the work that gets you recommended across all of them.

How AI search engines choose which businesses to recommend

An AI assistant answers a "who should I hire" question in two ways. It can draw on what the model absorbed in training, or it can search the web live, read a handful of pages and write an answer from them. For local and commercial questions in 2026, live search does most of the work: the engine runs several searches behind the scenes, pulls passages from the pages it finds, compares them, and names two or three businesses it can support with sources.

That has three consequences for a business owner:

• If the crawler cannot fetch your page, you are not a candidate. No amount of good content helps a page the engine never read.
• The engine trusts what it can confirm in more than one place. Your own site says you are the best roofer in Etobicoke. Reviews, directories, a local news mention and a "best of" list saying it too is what makes the claim usable.
• It quotes passages, not pages. A clear two-sentence answer near the top of a section is far more likely to be lifted than the same fact buried in paragraph six.

Each engine gets its live results from a different place, which is why the crawler question matters more than most people expect:

We group the work into four plays, the same ones we run inside every SEO programme: be legible, be quotable, be corroborated, be measured. The first step comes before all four.

Step 1: Let the AI crawlers in

This is the most common reason a good business is invisible to ChatGPT, and it is usually nobody's decision. It happens in three places.

Your robots.txt

Open yourdomain.ca/robots.txt in a browser. If you see a Disallow rule under OAI-SearchBot, PerplexityBot, Claude-SearchBot or a blanket rule for all user agents, the search features of those assistants are being told to stay out. Some security plugins and SEO plugins added blanket AI-bot blocks in 2024 and 2025 when the debate was about training data. A sensible 2026 position for most service businesses is: allow the search crawlers, and decide separately whether you care about the training crawlers. Those are different bots with different names, and you can treat them differently.

Your host, CDN or firewall

robots.txt is a request. A firewall is a wall. Cloudflare began offering one-click AI-crawler blocking in 2024 and moved to blocking AI crawlers by default on new domains in 2025, and many managed hosts run bot-protection rules that challenge anything that is not a mainstream browser. Your robots.txt can say "welcome" while your CDN returns a challenge page. Check your CDN's bot or AI crawler settings, and ask your host directly whether OAI-SearchBot and PerplexityBot receive a normal page. If your server logs are available, search them for those user agents and look at the status codes.

Pages that only exist in JavaScript

Many AI crawlers do not execute JavaScript the way Googlebot does. If your services, prices or locations only appear after a script runs, the crawler may see an empty shell. View the page source (not the inspector) and check that your key content is in the HTML. Server-rendered or pre-rendered pages are the safe default.

Step 2: Be legible, so engines know exactly who you are

AI engines think in entities: a specific business, at a specific address, offering specific services, in specific places. Your job is to remove every reason for the engine to be unsure which business it is looking at.

• One version of your facts. Business name, address, phone, hours and service area should be identical on your site, Google Business Profile, Apple Business Connect, Bing Places, Yelp, your industry directories and your social profiles. "Ste 400" in one place and "Suite 400" in another is minor; an old address on three directories is not.
• A real About page. Who owns the business, when it started, where it operates, who does the work, any licences or memberships. Engines lean on this when deciding whether you are a real operator or a lead-generation site.
• A page per service and per area you actually serve. "We do everything" pages give an engine nothing to match a specific question against.
• Organization or LocalBusiness schema with sameAs links to your Google Business Profile, LinkedIn, Instagram and the directories that list you. This ties the scattered profiles into one entity. The copy-paste templates are in the schema guide linked above.

A note on llms.txt: it is a proposed plain-text file that summarises your site for language models. We publish one on creativescope.ca because it takes twenty minutes. Treat it as a minor signal at best. None of the major engines has said it uses llms.txt to choose who to recommend, and it will not rescue a site that fails the steps above.

Step 3: Be corroborated by the sources AI cross-checks

This is the step that separates businesses that get named from businesses that merely exist. When an assistant recommends a Toronto accountant or a GTA renovation company, the sources it cites are rarely just the businesses' own websites. They are review platforms, directories, local news, "best of" roundups, Reddit threads and industry associations. If you are absent from those, the engine has nothing independent to point to.

• Reviews, with substance. Volume matters, but so does what the reviews say. A review that mentions the service and the neighbourhood ("replaced our furnace in Leslieville, same day") gives an engine a quotable fact. Ask for reviews in a way that invites detail. Our Google Maps guide covers the review system we install.
• The directories that matter in your category. Not five hundred citation-farm listings: the ten to twenty that a buyer in your industry would actually use, plus your chamber of commerce or industry association.
• Listicles and roundups. "Best [service] in Toronto" articles on other people's sites are heavily cited by AI answers. Find the ones that already rank, check whether they accept submissions or updates, and give the authors a reason to include you. A listicle you publish about yourself on your own blog carries far less weight.
• Press and expert mentions. A quote in a local publication, a podcast appearance, a guest article in a trade outlet. Unlinked brand mentions still count here, which is one way AIEO differs from classic link building.
• Community answers. Genuine participation in Reddit and industry forums. Engines quote these threads often. Astroturfing gets noticed by people first and engines second.

Corroboration compounds. In our DKI Canada work, the first job was making 70+ franchise locations that looked like seventy different companies read as one brand with consistent details, because a network nobody can verify is a network nobody recommends.

Step 4: Be quotable with answer-first content

An engine that is writing an answer looks for sentences it can lift. Make that easy.

• Open every service page with a direct answer: what you do, for whom, where, and roughly what it involves. One or two sentences, before the marketing copy.
• Turn your sales-call questions into headings. "How long does a kitchen renovation take in Toronto?" as an H2, answered in the first sentence beneath it.
• Use specific numbers with units. "Most installs take one day; larger homes take two" is quotable. "Fast, reliable service" is not.
• Publish comparison and "how to choose" content in your category, and be honest about when you are not the right fit. Balanced pages get cited; sales pages rarely do.
• Date and refresh key pages. Visible updated dates and current-year facts help with engines that favour recency, Perplexity in particular.

Step 5: Test your visibility with a fixed prompt set

AI answers vary from run to run and person to person, so one search proves nothing. A fixed prompt set, run on a schedule, does.

1. Write 15 to 25 prompts the way a buyer would ask them. Mix the types: "best [service] in [neighbourhood]", "who should I hire for [problem] in Toronto", "[your business] vs [competitor]", "is [your business] any good", and a few problem-first questions that never mention a service category.
2. Run each prompt in ChatGPT, Perplexity, Gemini and Google (for the AI Overview or AI Mode answer), logged out or in a fresh session, with location set to where your customers are.
3. Record three things per answer: whether you are named, who else is named, and which sources are cited.
4. Repeat monthly with the same prompts. Your number is the share of answers that name you. The cited sources tell you where to go and get corroborated next.

Paid tools now automate this across hundreds of prompts, and they are worth it at scale. For a single-location business, a spreadsheet and an hour a month is enough to see whether the work is moving anything.

Step 6: Measure AI referrals in GA4

Being named is the goal; visits and enquiries are the proof. In GA4, open Reports, then Acquisition, then Traffic acquisition, and set the dimension to session source. Look for chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai. ChatGPT also tags many of its outbound links with utm_source=chatgpt.com, which makes it easy to spot.

• Build a custom channel group called "AI assistants" that matches those sources, so the number sits next to Organic Search in every report instead of hiding inside Referral.
• Track conversions by that channel, not just sessions. AI referrals tend to be small in volume and high in intent.
• Add "ChatGPT or another AI assistant" to your "How did you hear about us?" field. Many AI-driven visits arrive as direct traffic because people copy the name into a browser or call from the app. The form field catches what analytics misses.

If your GA4 setup is shaky to begin with, fix that first; our GA4 setup guide covers consent mode and conversion tracking properly.

What not to waste money on

AI search has attracted the usual crowd of shortcuts. Skip these:

• Guaranteed ChatGPT placement. Nobody controls what an assistant says. Anyone promising a guaranteed recommendation is selling the same thing as guaranteed page-one rankings were in 2012.
• Mass AI-generated pages. Hundreds of thin location pages give engines nothing new to quote and can drag down how search engines see the whole site.
• Fake or incentivised reviews. They breach platform rules, and review text is exactly what assistants quote back to buyers.
• Treating llms.txt as a strategy. Publish it, then get back to the work that matters.

The honest summary: getting recommended by AI is mostly good SEO, done consistently, with more attention on corroboration and measurement than most programmes give it. Toronto is still early. In most categories we check, the businesses being named are not the best ones, just the most verifiable ones. If you would like help closing that gap, our SEO and AIEO programme runs all four plays under one retainer.