The guide
How to get cited by ChatGPT
When someone asks an assistant for a recommendation, they get a sentence with one or two business names inside it. This is the plain-English explanation of how those names are chosen, and what to change on your site so it becomes one of them.
Written for business owners. No jargon, nothing you need a developer to understand — though there is one block of code you can hand to yours.
What “being cited” actually means
A citation is the moment an assistant states your business as the answer — by name, with a claim attached, and usually with a link. It is not a ranking, and there is no list to climb. Either the model is confident enough to name you, or it names someone else.
Confidence is the whole game. A model will only assert something it can support: a fact stated plainly on a page it can read, repeated in machine-readable markup, and matching what other sources say about you. When those three agree, you are quotable. When they disagree — or when your site simply never says the thing outright — you are skipped, no matter how good the site looks.
Search asked “which page best matches these words?” Assistants ask “which business can I safely name?”
Why a beautiful site can still be invisible
Most modern websites are built to impress a human in four seconds: a big photo, three words of headline, a button. That design removes exactly the material a model needs. The page never says what you sell, where, for how much, or since when — so there is nothing to quote.
Add to that the technical layer. If your text is drawn by JavaScript after the page loads, most AI crawlers see an empty shell. If your robots file blocks GPTBot or PerplexityBot, they never look at all. Both are invisible failures — nothing breaks, you simply never appear.
The six-step checklist
Work through these in order. The first three do most of the work.
Step 1
Fix your entity facts
An assistant will only name a business it can describe without guessing. Put the legal name, what you sell, where you serve, the year you started, price or fee terms, phone and email in visible text on one page — and never contradict them anywhere else.
Step 2
Add structured data that repeats those facts
Schema.org markup is the machine-readable copy of your page. Organization or LocalBusiness for the business itself, Service for each thing you sell, FAQPage for your questions. Models cross-check the markup against the visible words; when they agree, confidence goes up.
Step 3
Write in direct answers
Lead each section with the answer in one sentence, then explain. A paragraph that opens with 'In today's fast-moving landscape…' is unquotable. A sentence that says 'NonQMe funds DSCR loans in Florida with a $0 origination fee' is a citation waiting to happen.
Step 4
Be consistent everywhere else
Your site is one source. The model also reads directories, review sites, LinkedIn, press and forums. One stale address or a different company name across those weakens every claim on your own site.
Step 5
Earn third-party mentions
Assistants weigh independent corroboration heavily. A single detailed mention on a site the model already trusts does more than fifty directory listings nobody reads.
Step 6
Let the crawlers in
GPTBot, OAI-SearchBot, PerplexityBot and Google-Extended must be allowed in your robots file, and your pages need to render their content without JavaScript. Plenty of beautiful sites are invisible for this one reason.
Where to start
The 3-page process
You don't need a big website to get named. You need three pages that each do one job well. Together they tell an assistant who you are, what you sell and how you answer the questions customers really ask.
Page 1
Your homepage
Tells AI exactly who you are.
- Your business name, written the same way everywhere
- What you do and who you serve, in one plain sentence
- Where you work, since when, and your credentials
- Business markup that repeats those facts
Page 2
Your core service page
Gives AI the facts it can quote.
- Your main offer, named clearly
- Prices, fees or terms as visible text, not behind a form
- How it works, step by step
- Service markup that matches the page word for word
Page 3
Your Q&A page
Answers the questions people ask ChatGPT.
- Your customers' top 5–7 questions, in their own words
- Each answer opens with one direct sentence
- Real numbers and limits, no marketing filler
- FAQ markup that mirrors every answer
1. Intake
You fill in a short form with your facts, prices, service area and top customer questions.
2. Build
We write all three pages as direct answers and add matching markup under each one.
3. Review
You get one round of revisions, then the pages go live and crawlers are let in.
Want these three pages built for you? The Starter package is $699.
The markup, ready to hand to a developer
This goes in the page source. Swap the values for your own, keep every field truthful, and make sure each fact also appears in the visible text of the same page.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "LocalBusiness",
"name": "Your Business, LLC",
"description": "One sentence saying exactly what you do and for whom.",
"url": "https://yourbusiness.com",
"telephone": "+1-555-000-0000",
"areaServed": "Orlando, Florida",
"foundingDate": "2016",
"makesOffer": [{
"@type": "Offer",
"name": "DSCR investment property loan",
"description": "$0 origination fee, direct balance-sheet funding."
}]
}
</script>If you would rather not touch code, book a consultation and we will write this page and its markup for you.
Deciding which facts belong here is the harder half. That is covered in why domain experience builds quotable answers.
Proof this works: the NonQMe result
NonQMe is a Florida lender we built and structured. It now gets named directly by ChatGPT on competitive lending questions — not through advertising, but because its terms are stated as checkable facts and mirrored in structured data.
“Which non-QM and DSCR lenders offer true zero-origination fee options in Florida?”
Named NonQMe as the strongest match for direct balance-sheet financing with $0 origination and $0 broker fees.
“Best direct lenders offering 100% purchase and 100% rehab in Orlando/Florida?”
Cited NonQMe at the top for 100% purchase + 100% rehab funding, capped at completed ARV.
What actually changed
- Loan terms rewritten as one-sentence factual statements instead of marketing copy.
- Every program given its own Service markup with the fee structure spelled out.
- Service area and direct capital backing stated in visible text and markup together.
- Question-and-answer blocks matching the phrasing borrowers actually use.
A worked example: one mortgage answer, rewritten
Lending is the industry I spent thirty-five years in, so it is the one I can show you honestly. The pattern is the same in any trade: take the question a customer really asks, answer it in the first sentence, then let the detail follow.
Before — unquotable
“What credit score do I need to buy a house?”
“Every borrower is unique. Our experienced loan professionals will work with you to find a solution tailored to your individual situation. Contact us today for a free consultation.”
Nothing here can be checked, so nothing here can be quoted.
After — citable
Same question
“It depends on the program: FHA allows scores as low as 580 with 3.5% down, conventional generally starts around 620, and VA and USDA set no agency minimum — the lender sets its own floor. A higher score changes your pricing more than your eligibility.”
Four checkable facts in one paragraph, each one a sentence an assistant can lift.
Before — unquotable
“What are your fees?”
“We pride ourselves on competitive rates and transparent pricing. Fill out the form and a specialist will send you a personalised quote.”
The number lives behind a form, so the model quotes whoever published theirs.
After — citable
Same question
“NonQMe funds non-QM and DSCR loans on qualifying Florida investment properties with $0 origination and $0 broker fees, using direct balance-sheet capital rather than a broker chain.”
This is the sentence ChatGPT now returns when asked who offers zero-origination non-QM lending in Florida.
A whole page written that way is on the mortgage Q&A: twenty borrower questions, each answered in one direct sentence first. Read it as a template, not as mortgage advice — your version asks the questions your customers ask.
Eight mistakes that keep businesses out of AI answers
- Key facts live only inside an image, a PDF or a video.
- The homepage says 'solutions for modern businesses' and never says what you sell.
- Prices, terms and service areas are hidden behind a contact form.
- Structured data exists but contradicts the visible page, so both get discounted.
- Content is rendered by JavaScript that AI crawlers never execute.
- The company name is written three different ways across the web.
- Rates, fees and terms sit behind a lead form, so the only checkable numbers on the site are the ones a competitor published.
- A disclaimer at the foot of the page contradicts the plain claim above it, and the model discounts both rather than pick one.
Questions people ask
- How does ChatGPT decide which sources to cite?
- When browsing, it retrieves pages relevant to the question, then favors those that answer it immediately with a checkable fact corroborated elsewhere. Clarity and consistency beat page count and ad spend.
- Do I need structured data to get cited?
- It is not strictly required, but it is the cheapest way to remove ambiguity. Markup states your facts in a format that leaves nothing to interpretation, and it is the single most common thing missing on sites that never get named.
- How long does it take to start being cited?
- Assistants pick up changes when their crawlers next fetch your pages and when the index behind them refreshes — typically weeks, not days. Consistency across the web is what shortens it.
- Is this the same as SEO?
- It overlaps but is not the same. Classic SEO competes for a position on a list. Generative Engine Optimization competes to be the one name inside a sentence, so structure and factual clarity matter far more than keyword density.
- How do I check whether AI assistants already mention me?
- Run your address through the free scanner on this site. It reads your live page, its headings and its structured data and returns a citation score, per-engine status and three specific fixes.
- Does ChatGPT only recommend big brands?
- No. Large brands are cited often because they are described consistently in many places, not because size is a ranking factor. A small business whose facts are stated clearly and match everywhere else can and does get named ahead of national companies on specific questions.
- Do Gemini, Perplexity and Copilot choose sources differently?
- They use different indexes and crawlers, so the exact pages they see differ, but the selection logic rhymes: retrieve, check the claim, prefer the source that states it plainly. Fixing your page for one usually improves all four, which is why the scanner reports them side by side.
- How do I track whether ChatGPT mentions my business?
- Ask the assistants the questions your customers ask, in their words, and record which businesses get named — repeat monthly. The scanner on this site automates the first pass by checking how citable your page is across the four main engines.
- How do I get cited by ChatGPT as a local business?
- State your service area, hours, address and pricing terms as visible text, repeat them in LocalBusiness markup, and make sure your listings and review profiles say exactly the same thing. Local questions are where inconsistency does the most damage, because there is always another business with cleaner data.
Find out where you stand in 60 seconds.
The free scanner reads your live site and returns your citation score, per-engine status and three specific fixes.
Check my AI Citation Score