How to get your business cited by ChatGPT and Perplexity

When someone asks an AI assistant for a recommendation in your category, you're either in the answer or you don't exist. There's no page two.

Here's what genuinely influences whether you're the source that gets quoted.

Understand how the answer gets built

Most AI assistants answering a current question do some form of retrieval: they search, pull a handful of sources, and synthesise. Some also draw on training data, which you can't influence directly.

The retrieval part is where you have leverage. It behaves more like search than like magic — which is why traditional SEO fundamentals still matter.

What increases your chances

1. Be findable in conventional search. Retrieval frequently runs through a search index. If you don't rank for the query, you're not in the candidate set. This is the least glamorous and most important point.

2. Write extractable statements. Models lift sentences. Make sentences that survive being lifted:

"A typical Google Ads management retainer in South Africa ranges from R6,000 to R25,000 per month depending on spend and scope."

That's quotable. "Pricing varies depending on your needs — get in touch for a quote" is not, and it never gets cited.

3. Answer the actual question in the actual phrasing. People ask "how much does it cost", not "what is our pricing philosophy". Match the language of the query.

4. Be specific with numbers, names and dates. Specificity signals authority to a model deciding between sources. Vague content loses to precise content every time.

5. Establish entity clarity. Structured data, consistent business information everywhere, sameAs links connecting your profiles. A model needs confidence about who you are before attributing a claim to you.

6. Get mentioned elsewhere. Being referenced on other credible sites influences whether you're treated as a source. This is the hardest and probably the most durable factor.

What doesn't work

Keyword stuffing. It didn't work for search and it works less here.

Writing for models instead of people. Content that reads as machine-targeted tends to be thin, and thin content doesn't get cited.

"AI SEO tools" promising placement. Nobody controls what a model cites. Anyone selling guaranteed AI visibility is selling a guess.

What I honestly can't tell you

Whether any single one of these moves the needle, with data.

Measuring AI citation is genuinely hard. There's no reporting interface. You can manually query assistants and note whether you appear, but that's a sample, not a measurement, and results vary between sessions and models.

I'm implementing all of the above on this site. I don't have clean attribution data proving it works, and I'd distrust anyone who claims they do this early.

The practical test

Ask ChatGPT, Perplexity and Google's AI Overview the ten questions your customers actually ask. Note who gets cited.

Then read those sources. In my experience they're usually specific, well-structured, and answer the question directly in the first paragraph — rather than being the most beautifully designed or most heavily marketed sites in the category.

That's the whole strategy: be the clearest, most specific, most easily quoted answer to a question people are genuinely asking.

Related: AI SEO vs traditional SEO and schema markup that actually matters.