How to Get Your Business Cited by ChatGPT and Perplexity

A meaningful share of the questions that used to begin as searches now end inside an AI assistant, with no click at all. Whether your business appears in that answer is decided by different mechanics than blue-link ranking, and the differences are learnable.

Cited is not the same as ranked

Traditional search rewards pages that rank. AI answers reward passages that can be extracted. A page sitting at position eight can be cited ahead of position one if its content is cleanly structured and the top result buries its answer under three paragraphs of throat-clearing.

This is genuinely good news for smaller businesses. You cannot out-authority a national competitor quickly, but you can out-structure them this afternoon.

The seven things that actually move it

1. Answer in the first forty words

Every important page should open with a direct, standalone answer to the question it targets. Standalone is the operative word — it must make sense lifted out of the page entirely, because that is exactly what happens to it.

2. Publish real numbers

Assistants prefer specifics because specifics are checkable. A page saying pricing is bespoke gives a model nothing to quote; a page saying websites start at $250 gives it a sentence. If your prices are hidden behind a contact form, you are invisible to every buyer whose research happens inside an assistant.

3. Use headings that match how people ask

How much does a website cost is a heading. Our pricing philosophy is not. Retrieval matches the shape of the question, so write headings in that shape.

4. Comparison content earns disproportionate citations

Balanced comparisons are among the most-cited formats, because a model answering which is better needs a source that discusses both. Comparisons that only flatter the author get skipped — including the cases where a competitor is the better choice is what makes the page usable, and usable is what gets cited.

5. Add structured data

FAQ, Organisation, Product and Article schema tell a machine what your content is rather than making it infer. Google states this is not required for its AI features, but other engines parse it readily and it costs nothing but attention.

6. Publish machine-readable files

An llms.txt summarising what you do and linking key pages, and a pricing.md listing prices in plain text, are trivially parseable. Autonomous agents comparing suppliers on a buyer's behalf can read them without rendering anything.

7. Be present where the models already look

Assistants cite third-party sources far more readily than a company's own site. Industry directories, review platforms, forum discussions and reputable roundups carry more weight per mention than your homepage does. This is the slowest item on the list and usually the most valuable.

Check your robots file before anything else

None of the above matters if the crawlers are blocked. GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot and Google-Extended must be allowed, and plenty of sites block them by accident through an over-broad rule or a security plugin's default.

What does not work

• Keyword stuffing, which measurably reduces AI visibility rather than merely failing to help.
• Writing separate thin pages aimed at machines, which lands squarely inside scaled-content-abuse territory.
• Chopping pages into fragments for easier extraction — normal headings and paragraphs already do this.
• Vague superlatives. Best-in-class is unquotable; a specific number with a date is quotable.

Measuring it

There is no console for this yet. The workable method is manual and takes fifteen minutes a month: pick your twenty most commercially important questions, ask them in a few assistants, and record whether you appear and who does instead. Month over month, that table tells you more than any dashboard currently on sale.