If someone asked ChatGPT right now to recommend a business like yours, would you appear in the answer?
For most small business owners and consultants, the answer is no — not because their business isn't good enough, but because their website isn't structured in a way that lets AI systems extract and cite it.
This guide covers the 7 fixes that matter most. Each one is explained in plain English: what it is, why it matters, and exactly how to do it. No developer required for most of them.
Fix 1: Check Your robots.txt for AI Crawlers
What it is: A file at /robots.txt that tells search bots which pages they're allowed to visit.
Why it matters: By default, many robots.txt files were written before AI search existed. They allow Google but may be accidentally ambiguous — or even blocking — the AI crawlers that feed ChatGPT, Perplexity, and Apple Intelligence.
How to do it: Add these lines to your robots.txt file:
User-agent: GPTBot Allow: / User-agent: OAI-SearchBot Allow: / User-agent: ClaudeBot Allow: / User-agent: PerplexityBot Allow: / User-agent: Google-Extended Allow: / User-agent: Applebot-Extended Allow: /
That's it. Save and deploy. You've just opened the door to 6 major AI crawlers.
Fix 2: Create a llms.txt File
What it is: A plain-text file at /llms.txt — essentially a brief that tells AI systems what your business does, who it's for, and what pages matter.
Why it matters: AI systems don't just crawl pages individually — they try to understand what a site is about as a whole. Without a llms.txt, they have to piece this together from multiple pages, and they often get it wrong or miss things entirely. This is the single highest-impact fix if you have nothing else in place.
How to do it: Create a file at /llms.txt with content like this:
# [Your Business Name] > [One sentence describing what you do] ## About [2-3 sentences. Who you help, what you do, your background.] ## Services - [Service 1]: [Description + starting price] - [Service 2]: [Description + starting price] ## Key Facts - Typical engagement: [timeline] - Pricing model: [fixed project / hourly / retainer] - Location: [your city] - Works with: [types of clients]
Keep it factual and specific. AI systems will use this as a primary source when forming answers about your business.
Fix 3: Add FAQPage Schema
What it is: A block of structured code (JSON-LD) added to your service pages that marks up your Q&A content so AI systems can extract it directly.
Why it matters: When someone asks ChatGPT "what does [type of service] cost?" or "how long does [service] take?", the AI is looking for pages that answer those questions clearly. FAQPage schema is the most direct signal: it literally says "here is a question, here is the answer." Research from Princeton's KDD 2024 paper found that adding specific statistics and authoritative Q&A formats increased AI citation rates by 80–115%.
How to do it: Add this to the <head> of your service pages, filling in your actual content:
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "What does [your service] cost?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Starting from $[price]. Most projects are [range]."
}
},
{
"@type": "Question",
"name": "How long does [your service] take?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Typically [X] weeks from kickoff to delivery."
}
}
]
}
</script>
Aim for 4–6 questions per page. Focus on what a prospect would actually ask before hiring you.
Fix 4: Add Review and AggregateRating Schema
What it is: Structured markup that turns your text testimonials into machine-readable reviews with a rating score AI can cite.
Why it matters: If you have client testimonials on your site, AI systems probably can't find them in a useful form. Plain text scattered through a page doesn't tell a crawler "this is a review, this is the rating." Schema markup does. When AI answers "is [business] reliable?", structured reviews are what it cites.
How to do it: Add to your homepage or service pages:
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "LocalBusiness",
"name": "[Your Business]",
"aggregateRating": {
"@type": "AggregateRating",
"ratingValue": "4.9",
"reviewCount": "12"
},
"review": [
{
"@type": "Review",
"author": { "@type": "Person", "name": "Client Name" },
"reviewRating": { "@type": "Rating", "ratingValue": "5" },
"reviewBody": "[Client quote]"
}
]
}
</script>
Fix 5: Make Your Content Extractable
What it is: Rewriting vague marketing copy into specific, citable facts.
Why it matters: AI citation works differently from Google ranking. Google rewards authority and links. AI systems reward specificity — they need something concrete to quote. "We deliver results" can't be cited. "We typically reduce manual ticket routing by 60–80% within 3 weeks" can be.
How to do it: Go through each service page and ask: what would a journalist need to know to write a sentence about this service? Then write that sentence on the page.
Replace:
"We help businesses streamline their operations."
With:
"We build Slack bots that route support tickets to Jira automatically. Clients typically save 8–12 hours per week in manual ticket handling. Delivery in 2–3 weeks. Starting from $5,000."
The second version has five citable facts. The first has zero.
Fix 6: Add Freshness Signals
What it is: HTML elements that tell AI crawlers when content was created or updated.
Why it matters: AI systems prefer recent sources when answering time-sensitive questions. A page with no date is treated as potentially old — which reduces the likelihood of citation. This is easy to fix and often overlooked.
How to do it: Add a <time> element to your service pages and homepage:
<time datetime="2026-03-17">Updated March 2026</time>
Add this visibly on the page — in the hero section or near the top of the content. Make a habit of updating it when you make significant changes to a page (quarterly is enough).
Fix 7: Test Visibility Before and After
What it is: A simple self-audit you can do in 10 minutes using free tools.
Why it matters: You can't optimize what you don't measure. Testing before you implement any fixes gives you a baseline. Testing after tells you whether the changes worked. Without this, you're guessing.
How to do it:
- Go to Perplexity.ai and search: "best [your service category] in [your city]". Write down what comes up.
- Try ChatGPT with: "I need a [your service] for my [type of business]. Who should I hire?"
- Try 3–5 variations of how your clients would describe their problem.
- Implement the fixes above.
- Wait 3–4 weeks (crawlers need to re-index).
- Repeat the same searches. Compare.
This is also useful before implementing fixes — you'll often discover which competitors ARE being cited, which tells you exactly what structural signals they have that you're missing.
How Long Does This Actually Take?
Here's a realistic time estimate if you're doing this yourself:
The hardest part isn't technical — it's writing the specific, citable content for Fix 5. That's where most business owners get stuck. The honest answer to "what do we actually deliver?" takes thought to write well.
Everything else is structure. And structure is fast.
