ChatGPT recommends businesses where three signals are clean: a clear identity on your own site, outside confirmation from reviews and directories, and extractable content the engine can lift. The Identity-Confirmation-Extraction Method gives you the order to build them in.
You typed your category into ChatGPT, the way a customer would. The answer named two businesses with confidence. Yours was not one of them.
That is a different kind of quiet than a Google page-two ranking. Google at least shows you somewhere. ChatGPT just picks, and if it has not picked you, it is usually for one of three structural reasons, none of them random and none of them about how good your service actually is.
The reason most owners get stuck is that they keep treating this as an SEO problem. It is not. There is no results page to climb. ChatGPT writes one paragraph and names one or two businesses inside it. You are either in that paragraph or you are not. Building for that is a different job, and it has its own three signals.
This post walks you through what those three signals are, why ChatGPT skips businesses that miss any one of them, and a 30-day plan to fix all three. By the end you will know which signal is weakest for you and what to change this week.
Key Takeaways
- ChatGPT does not rank pages. It picks one or two businesses to name in a synthesised answer, so the work is recommendation engineering, not link-building.
- Three signals decide whether you get named: a clear identity on your homepage, outside confirmation through Google Reviews and directories, and extractable content blocks the engine can lift directly.
- A complete Google Business Profile correlates with 2.8 times higher AI search appearance, and Google Reviews are one of the strongest fast-fix trust signals for AI visibility.
- Princeton GEO research found that adding statistics improves AI visibility by roughly 41 percent, expert quotations by 28 percent, and citations by 30 percent. Specificity beats volume.
- The Identity-Confirmation-Extraction Method orders the work: define the entity, prove it exists outside your own site, then format the answer ChatGPT will quote back.
Why does ranking on Google not get you into ChatGPT answers?
Because ChatGPT is not running a ranking algorithm. It is running an answer generator, and answer generators behave differently from search engines.
A search engine returns a list of links and lets the human decide. The work for the business is to climb the list. An answer generator returns one synthesised paragraph and names one, two, sometimes three businesses inside it. The human sees the answer, not the list. The work for the business is to be the option the engine names. There is no second page to recover from.
That difference changes everything about what counts as a strong signal. A high Google ranking comes from backlinks, keyword relevance, and domain authority. A ChatGPT recommendation comes from whether the engine can identify your business clearly, verify it through outside sources, and extract a confident answer about you from your own content. Some inputs overlap. Most do not.
The clearest example is local recommendations. Research into ChatGPT's local recommendation behaviour shows that Foursquare contributes a much larger share of the data signal than Google Maps for non-Google AI engines. A business with a strong Google Business Profile but no Foursquare or Yelp presence may sit happily in Google's local pack and stay invisible to ChatGPT for the same query.
The second failure mode is content shape. ChatGPT does not rank your page; it lifts text from it. A page with no FAQ structure, no schema markup, and no clear question-and-answer blocks gives the engine nothing reliable to lift. It moves on to a competitor whose page is shaped to be quoted.
So what for you: the moment you stop trying to rank and start trying to be the recommendation, the work changes shape. The three signals below are what that work looks like.
What does the research say ChatGPT actually uses to decide who to recommend?
The most useful evidence sits in three studies, each looking at the same problem from a different angle.
SE Ranking analysed 2.3 million pages and found that domain traffic is the single strongest predictor of AI Mode citations. High-traffic sites earn roughly three times more AI citations than low-traffic sites. Traffic is not the cause; it is the proxy for the underlying signals AI engines actually evaluate, including content quality, brand recognition, and review presence. The same study found that Google Business Profile completeness correlates with 2.8 times higher AI search appearance. (SE Ranking, 2026.)
Princeton University's Generative Engine Optimisation research, published at ACM KDD 2024, broke down what made content more visible to AI engines once it was already on the page. Adding statistics improved visibility by approximately 41 percent. Adding expert quotations improved it by around 28 percent. Citing authoritative sources lifted it by roughly 30 percent. Structured heading hierarchies appeared in 68.7 percent of cited pages, and structured data appeared in 61 percent. (Princeton GEO, 2024.)
Independent analysis of ChatGPT's local recommendation patterns identified three pathways the engine uses to verify a business: directory and aggregator presence (Foursquare, Yelp, industry-specific listings), review platform signals (Google Reviews, Trustpilot, sector-specific platforms), and the business's own structured site content. Businesses appearing consistently in ChatGPT recommendations had clear signal on at least two of the three. (Search Engine Land, 2026.)
Pull all three together and a pattern shows up. ChatGPT does not look for a single trust score. It runs three quiet checks: can it identify the business, can it verify the business outside the business's own site, and can it extract a clean answer about the business. If any one of those checks fails, the recommendation goes elsewhere.
So what for you: this is not seven things. It is three. And the three have a build order.
The Identity-Confirmation-Extraction Method
The Identity-Confirmation-Extraction Method is the working model for getting named in ChatGPT answers. It puts the three structural checks ChatGPT runs into the order you should build them, with one specific change per signal. Use it when you are starting from invisibility, or when you are trying to figure out why a strong-on-paper business is not converting into ChatGPT mentions.
Signal 1: Identity. Can ChatGPT tell, in one sentence, who you are, what you do, who you serve, and where you work?
The headline at the top of your homepage is the first thing ChatGPT reads when it tries to identify a business. A vague headline like "Your trusted local partner" gives the engine nothing to work with. A specific headline like "Smith & Co, employment lawyers in Melbourne specialising in unfair dismissal claims for small business owners" gives it everything: name, profession, location, service, customer. This is the entity definition problem, and it is the most commonly missed of the three signals. Action: rewrite your H1 and opening paragraph this week so it answers all four questions in plain language.
Signal 2: Confirmation. Do credible outside sources confirm your business exists, beyond your own website?
ChatGPT will not recommend a business it can only verify through the business's own claims about itself. It needs corroboration. Three things carry most of the weight here. A complete, recently-updated Google Business Profile, because GBP completeness correlates with 2.8 times higher AI search appearance and Google's own AI engines read it directly. A working stream of recent Google Reviews, because Google Reviews are one of the strongest fast-fix trust signals AI engines weight, and recent ones land inside the freshness window AI rewards. And presence on at least one third-party platform relevant to your sector, whether that is Trustpilot, Yelp, Foursquare, or a sector-specific directory. Action: complete your Google Business Profile fully this week, set up a workflow to ask for one new Google Review per week, and claim or create one third-party profile this month.
Signal 3: Extraction. Is your content shaped so ChatGPT can lift a clean answer from it?
ChatGPT generates answers by extracting from content shaped like answers. Three things matter here. Question-led headings written the way a customer types the query, because AI engines lift answers from question-shaped headings. FAQ blocks of five to ten customer-language questions answered directly underneath, because FAQ pages are the single most cited content format on AI engines. And schema markup, because schema is foundational and missing schema is a measurable visibility deduction. Princeton's GEO research found structured data appeared in 61 percent of AI-cited pages. Action: write five to ten FAQ entries in customer language this fortnight, ask your developer to add Organisation, LocalBusiness, Service, and FAQPage schema, and convert your top three section headings into questions.
The signals stack in that order for a reason. Identity is the gate: if ChatGPT cannot tell who you are, the other two do not get evaluated. Confirmation is the trust layer: it tells the engine your identity claim is real. Extraction is the conversion layer: it gives the engine the actual sentences to quote when it names you.
So what for you: most businesses missing from ChatGPT have one of these three weakened or missing. Diagnose which one is yours before you start building.
How do you build the three signals over the next 30 days?
A 30-day plan that maps each signal to a specific action, with time estimates per step.
-
Day 1 to 3. Audit your identity signal. Type your business name into ChatGPT. Then type your category and location ("best accounting firm in Brisbane", "employment lawyer for small business in Melbourne"). Note what comes back, who gets named, and whether the engine seems to know what you do. Read your own homepage H1 and ask yourself whether it answers four questions: who you are, what you do, who you serve, and where. If the answer is no on any one, identity is your weakest signal.
-
Day 4 to 7. Rewrite your homepage and tighten your headline. Replace any vague headline with one that names all four. "Welcome to Smith & Co" becomes "Smith & Co, employment lawyers in Melbourne specialising in unfair dismissal claims for small business owners." Keep the opening paragraph in the same plain language. This single change reshapes how every AI engine reads your site for the next year.
-
Day 8 to 14. Complete your Google Business Profile fully. Hours, services, photos, business description, products, attributes, and primary category. GBP completeness is one of the strongest trust signals for AI visibility, and the gap between half-complete and fully-complete profiles is where most owners lose AI recommendations. Set up a recurring weekly review-request habit so one new Google Review lands per week, sent within seven days of the service. This is the lowest-effort change with the longest tail.
-
Day 15 to 21. Write your FAQ block and add it to the site. Five to ten questions in customer language ("how much does an unfair dismissal claim cost?" not "our pricing structure"). Direct, self-contained answers underneath. Publish on a single FAQ page or alongside the relevant service page. While you are at it, convert your top three section headings into questions. This makes your content extractable, which is what ChatGPT needs to quote you.
-
Day 22 to 27. Add schema markup and an llms.txt file. Ask your developer (or your CMS plugin) to add Organisation schema, LocalBusiness schema if location matters, Service schema for your main services, and FAQPage schema on your new FAQ block. Write a short llms.txt file listing your key pages with one-line descriptions and upload it to your site root. The format is at llmstxt.org and the whole thing takes about 30 minutes once the page list is decided.
-
Day 28 to 30. Claim one third-party profile and run a free AEO scan. Pick the directory or review platform most relevant to your sector and create a complete profile. Then run the scan to see what changed. The scan reads your Google Business Profile, your schema, your reviews, and your site content, and gives you a visibility score across four pillars: Technical Health, Content Quality, Authority and Trust, and AI Presence. The gap between your starting position and your day-30 score tells you which signal moved the most.
You do not need all three signals at full strength in 30 days. You need the weakest one identified and the gap closed enough for ChatGPT to find you. The lift compounds from there.
Frequently asked questions
How long does it take to start ranking in ChatGPT answers?
Mechanical fixes such as schema markup, an llms.txt file, and a FAQ section start showing up within days to a few weeks. Trust signals like Google Reviews, Google Business Profile completeness, and third-party mentions take weeks to months to compound. Most businesses see measurable improvement in ChatGPT recommendations within 30 to 60 days of fixing the top two or three signals.
Does ChatGPT actually use my Google Reviews when it recommends businesses?
Yes. Google's own AI engines including AI Overviews, Gemini, and Maps read Google Reviews directly. ChatGPT and other third-party AI tools use cached snippets, indexed aggregator content, and pattern extraction across the web. Reviews reach AI through multiple pathways. The GetRecommended.io scan flags Google Reviews as a key trust signal because the data backs the mechanism. The deeper read on this is in how AI search engines actually use your Google Reviews.
If I rank well on Google, will I automatically rank in ChatGPT answers?
Not reliably. Google rankings and ChatGPT recommendations share some inputs such as content quality and structured data, but ChatGPT also draws heavily from sources Google does not rank, including Foursquare, Yelp, Trustpilot, industry directories, and review pattern data. A page-one Google ranking does not guarantee ChatGPT recommendations. A modest Google presence with strong identity and outside confirmation often beats it.
What is the single fastest fix to improve my chances of being named by ChatGPT?
Rewrite your homepage headline so it states who you are, what you do, who you serve, and where, in plain language. ChatGPT reads the H1 and opening paragraph first when it tries to identify a business. A vague headline gives it nothing to recommend. A specific one gives it everything. The change costs nothing and often moves visibility within a fortnight.
Do I need schema markup for ChatGPT to recommend my business?
Yes, in practice. Schema markup is foundational, and missing schema is a measurable visibility deduction in the GetRecommended.io scan. Princeton GEO research found structured data appears in 61 percent of AI-cited pages. Organisation, LocalBusiness, Service, and FAQPage schema together give ChatGPT a verified anchor point. Without them the engine makes inferences from unstructured text and tends to be conservative.
The Bottom Line
ChatGPT is not running a ranking algorithm and there is no page two to recover to. It picks one or two businesses to name in a synthesised answer, and if it has not picked you, it is almost always because one of three signals is weak: identity, confirmation, or extraction.
The good news is that the three signals are ordered, and the gate is the cheapest one to fix. A clear homepage headline costs nothing. A complete Google Business Profile and a steady weekly review-request habit cost an hour to set up. FAQ structure and schema markup cost an afternoon. None of this is platform-paid. None of it requires a new agency.
Run a free AEO scan to see exactly which of the three signals is weakest for your business right now and what to fix first. Five minutes, no credit card, a clear next step. For a wider view of the full set of AI visibility signals, read how to get recommended by AI search engines. Common questions about scan results are answered on the scan FAQ.