Oct 22, 2026 · The Boost My AI Search Team

How to Get Your Business Recommended by ChatGPT

By The Boost My AI Search Team, Answer engine optimization (AEO) · October 2026

To get ChatGPT to recommend your business, you need five things in place: AI crawlers must be allowed to read your site, an llms.txt file must tell AI what you do, your service pages must answer questions directly, your schema markup must match what customers actually ask, and third-party sources must confirm your expertise. Fix all five and your recommendation rate rises measurably within 90 days.

Why does ChatGPT recommend some businesses and not others?

ChatGPT names businesses whose content it can read, understand, and trust. When a user asks "Who is a good family law attorney in Denver?" the model draws on text it has indexed from the web, third-party directories, and review platforms. If your site is blocked to AI crawlers, written in a way that buries the answer, or uncorroborated by outside sources, the model skips you — not because it dislikes you, but because it cannot confidently attribute a claim to you.

Three factors drive the decision: access (can the crawler read the page?), clarity (does the page answer the question directly?), and corroboration (do other credible sources agree?). SEO optimizes for click-through from a results page. Answer engine optimization (AEO) optimizes for being the answer itself. The two disciplines overlap but are not the same — see how AEO differs from SEO for a full breakdown.

Fix 1: Is your website visible to GPTBot and other AI crawlers?

The fastest way to disappear from ChatGPT recommendations is to block GPTBot in your robots.txt file — and many sites do this accidentally.

OpenAI's crawler is named GPTBot. Google's AI crawler is Google-Extended. Perplexity uses PerplexityBot. A blanket disallow rule like the one below blocks all of them:

User-agent: *
Disallow: /

So does a targeted block that was added during a security review and never revisited:

User-agent: GPTBot
Disallow: /

To allow AI crawlers while still blocking others, use explicit allow rules:

User-agent: GPTBot
Allow: /

User-agent: Google-Extended
Allow: /

User-agent: PerplexityBot
Allow: /

Check your live robots.txt at yourdomain.com/robots.txt right now. If you see a disallow for GPTBot or a wildcard block with no exceptions, that is the single highest-priority fix. Our free AI check tests this automatically and flags it in the report.

Fix 2: Do you have an llms.txt file that tells AI what you do?

An llms.txt file is a plain-text document at the root of your domain that tells AI systems — in structured, unambiguous language — who you are, what you do, and which pages matter most. Think of it as a cover letter written specifically for a language model rather than a human reader.

A minimal, effective llms.txt looks like this:

# Acme Plumbing — llms.txt
## Who we are
Acme Plumbing is a licensed plumbing contractor serving the Denver metro area since 2009.
We specialise in emergency pipe repair, water heater installation and drain cleaning.

## Key pages
- Services: https://acmeplumbing.com/services/
- Service area: https://acmeplumbing.com/service-area/
- Reviews: https://acmeplumbing.com/reviews/
- Contact: https://acmeplumbing.com/contact/

## What we do not do
We do not offer HVAC or electrical work.

Keep it factual and specific. Vague phrases like "we are passionate about excellence" carry no signal. Concrete facts — license numbers, service area zip codes, years in business, named specialties — are what a language model can actually use when deciding whether to name you. For a deeper guide on the format and who reads it, see our llms.txt explainer.

Fix 3: Are your service pages written in the answer-first format AI quotes?

AI engines extract passages, not pages. If the answer to a customer's question is buried in paragraph six after a history of your company and a stock-photo description, the model will not find it — or will find a competitor's page that leads with the answer.

Answer-first structure means the direct answer appears in the first one or two sentences of every section. Here is the same content written two ways:

Format	Opening of a "Do you offer same-day service?" section	AI extractable?
Traditional	"At Acme Plumbing, we have been serving Denver families for over a decade with a commitment to quality…"	No
Answer-first	"Yes. Acme Plumbing offers same-day emergency service across the Denver metro, available 7 days a week including holidays."	Yes

Every service page should answer at least three questions a customer would ask ChatGPT: what you do, who you serve, and what makes you different. Keep paragraphs short — one idea each. Use <h2> and <h3> headings that mirror the phrasing of real questions, not internal marketing labels. Our results page shows how recommendation rates change after service pages are rewritten this way.

Fix 4: Does your schema markup match the questions customers actually ask?

Schema markup is machine-readable metadata embedded in your HTML. It tells search engines and AI systems exactly what type of entity your page describes. Two schema types correlate most strongly with AI citations for small businesses:

• LocalBusiness (or a subtype like MedicalClinic, LegalService, HomeAndConstructionBusiness): Declares your name, address, phone, service area, hours and categories in a format AI can parse without reading prose.
• FAQPage: Pairs questions with answers in structured JSON-LD. When a customer asks ChatGPT a question that matches one of your FAQ entries, the model has a clean, attributable passage to cite.

A minimal FAQPage block in JSON-LD looks like this:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "Do you offer same-day plumbing repair in Denver?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Yes. Acme Plumbing provides same-day emergency repair across the Denver metro area, 7 days a week."
      }
    }
  ]
}
</script>

The questions in your schema should come from real customer queries — not from guessing. We track which questions customers are actually asking ChatGPT and Gemini about businesses like yours every week, which is how we know which FAQ entries are worth writing. See how we test AI answers for the methodology.

Fix 5: Are third-party sources (directories, reviews, press) confirming your expertise?

ChatGPT does not rely solely on your own website. It weights corroboration: if Google Business Profile, Yelp, Healthgrades, Avvo, Houzz, or industry press all say the same thing about your business — same name, same specialty, same location — the model gains confidence that you are a real, credible option worth naming.

Corroboration gaps that suppress recommendations include:

• Your Google Business Profile category does not match the specialty you describe on your website.
• Your business name is spelled differently across directories (e.g., "Acme Plumbing LLC" vs "Acme Plumbing Co.").
• You have no reviews on the platform most relevant to your industry (e.g., a clinic with no Healthgrades listing).
• Trade publications or local press have never mentioned you by name in a factual context.

The fix is systematic: audit your top five directory listings for name/address/phone consistency, claim any unclaimed profiles, and actively request reviews on the platform your customers use most. For B2B businesses, one well-placed mention in an industry newsletter or a contributed article in a trade publication carries more corroboration weight than dozens of generic directory entries.

How do you measure whether ChatGPT is naming you more often?

Measurement is the part most guides skip, and it matters because without a baseline you cannot tell whether your fixes are working.

The correct method is repeated, tracked asks: ask ChatGPT (and Gemini) the same set of customer questions every week, record whether your business is named, and calculate a recommendation rate over time. A single screenshot is not a measurement — AI answers vary by session, by phrasing, and over time as models update. A rate across 25 or 50 tracked questions, repeated weekly, is.

At Boost My AI Search, we run this process for every customer every week. A typical starting baseline for a business that has not done any AEO work is a recommendation rate under 10% across their tracked question set. After the five fixes above are in place — crawler access, llms.txt, answer-first pages, schema, and third-party corroboration — rates climb measurably within 60–90 days. We publish the methodology rather than just the outcome because the method is what makes the number trustworthy.

If you want to run a manual spot-check before committing to weekly tracking, ask ChatGPT: "Who is a good [your service] in [your city]?" Ask it five times in separate sessions and note how often you appear. That gives you a rough baseline. For a structured baseline across 25 real customer questions, the free AI check does it automatically.

The five fixes at a glance

Fix	What to check	Priority
AI crawler access	robots.txt allows GPTBot, Google-Extended, PerplexityBot	Highest — fix first
llms.txt	Plain-text file at root with factual description and key page URLs	High
Answer-first pages	Every service page leads with a direct answer, not a brand story	High
Schema markup	LocalBusiness and FAQPage JSON-LD matching real customer questions	Medium-high
Third-party corroboration	Consistent name/specialty/location across directories, reviews, press	Medium — ongoing

Frequently asked questions

How long does it take for ChatGPT to start recommending my business after I make these fixes?

Most businesses see a measurable increase in recommendation rate within 60–90 days of fixing crawler access, rewriting service pages, and adding schema. The exact timeline depends on how often AI models refresh their training data and how competitive your category is. Fixes to robots.txt and llms.txt can take effect faster because they affect live web crawls rather than training data alone.

Does having a Google Business Profile help ChatGPT recommend me?

Yes, indirectly. ChatGPT draws on information indexed from across the web, including data that originates from Google's index. A complete, accurate Google Business Profile — correct category, consistent name, recent reviews — adds corroborating signals that make the model more confident naming you. It is one piece of the corroboration fix, not a standalone solution.

Can I block some AI crawlers but allow others?

Yes. Your robots.txt can allow GPTBot and block others, or allow all AI crawlers selectively. The key is to be explicit: a wildcard disallow blocks everything unless you add named exceptions. If you are concerned about content scraping, allow the crawlers that power recommendation engines (GPTBot, Google-Extended, PerplexityBot) and block others by name.

What is the difference between AEO and SEO for a small business?

SEO optimizes for ranking in a list of links on a search results page. AEO optimizes for being named directly in an AI-generated answer. SEO measures clicks and rankings. AEO measures recommendation rate — how often the AI names you when a customer asks a relevant question. Both matter, but the tactics differ: AEO prioritizes answer-first structure, schema, and corroboration over link building and keyword density.

Do I need to do all five fixes, or will one or two be enough?

Fixes one and two (crawler access and llms.txt) are table stakes — without them, the others cannot work. Beyond that, the fixes compound: answer-first pages without schema are less citable than pages with both. Third-party corroboration amplifies everything else. Doing all five produces significantly better results than any single fix in isolation.

Want to know whether ChatGPT and Gemini recommend you today? Run the free AI check. If you would rather have it fixed every week, Boost My AI Search starts at $149 a month, month to month.

Last updated: October 2026