Essay · 06

How to get your company cited by AI search engines like ChatGPT and Perplexity

4 July 2026 · Updated 18 July 2026 · 7 min read

To improve your company's chances of being cited by AI search engines like ChatGPT and Perplexity, four things matter: let the AI crawlers read your site (check robots.txt and add JSON-LD structured data), write pages that answer real buyer questions directly enough to be quoted, build your presence on the high-authority third-party sites those engines actually cite (Reddit, YouTube, comparison sites, industry press), and measure your citation rate so you can see what changes. AI engines rarely cite a company from its own marketing pages alone. They cite the sources that already talk about you, so eligibility means being both readable and independently talked-about. There is no submission form and no paid placement; the engines decide what they cite.

Here's the deal: this is a different job from ranking on Google, and most advice treats it like the same job with new keywords. It isn't. Below are the four levers, in the order I'd work them.

Why AI citations matter now

Traditional search is shrinking. Gartner predicts search engine volume will drop 25% by 2026 as people ask AI assistants instead (Gartner, 19 February 2024). And when an AI answer appears, it keeps the click: Pew Research Center found Google users clicked a traditional result link in only 8% of visits where an AI summary showed, versus 15% where it didn't — and clicked a source cited inside the summary just 1% of the time (Pew Research Center, 22 July 2025). The takeaway is blunt: if the AI answer doesn't mention you, most buyers never reach your site to learn you existed. Being in the answer is the new being on page one.

Make sure the AI crawlers can read your site

Before an engine can cite you, it has to be allowed to read you — and that permission lives in a file most owners have never opened. Check two things. First, robots.txt: type yourcompany.com/robots.txt into a browser and look for Disallow rules under agents like GPTBot, ClaudeBot, Google-Extended, PerplexityBot, or OAI-SearchBot. Blocking one doesn't block the others — and a CDN can add these blocks without telling you, so check the raw file, not the rendered page. Second, structured data: add JSON-LD schema.org markup — Organization on your homepage, Article or BlogPosting on posts, FAQPage on any page with a real FAQ — so the engine can parse what your business is, who wrote a page, and when. Schema must mirror what a human sees: invisible FAQ markup with no visible FAQ reads as cloaking to the same engines you're courting, and hurts you.

Write pages that are shaped to be quoted

AI engines lift self-contained chunks of text, so the unit of work is the paragraph, not the page. Each answer should stand on its own: lead with the direct answer in a sentence or two, then expand, and never rely on "as mentioned above" or an orphan "we" that only makes sense with the rest of the page around it. Phrase your headings the way a buyer would ask an AI engine a question, then answer that exact question underneath. Define your entities plainly — what your product is, how it differs from the obvious alternative — because engines lift clean definitions. Back claims with specifics that are safe to quote: a real statistic with a named source and a year, a concrete number, a dated example. Generic marketing copy is uncitable because there's nothing in it worth lifting. If a paragraph wouldn't survive being pasted into an answer with zero surrounding context, rewrite it.

Get onto the third-party surfaces AI engines actually cite

This is the lever most companies miss. When an AI engine recommends a company, it usually cites a third party rather than that company's own site. Muck Rack's Generative Pulse study found earned media — journalism, reviews, and independent coverage — accounts for around 84% of AI citations, a figure held between 82% and 89% since mid-2025 (Muck Rack Generative Pulse, May 2026). In plain terms: the competitor gets the recommendation, but a Reddit thread, a review site, or an industry article gets the link. So the work is to be genuinely present and well-regarded on the crawlable, high-authority surfaces these engines lean on — Reddit, YouTube, comparison and review sites like G2, and earned press in your industry. This is earned, not bought or gamed: real reviews from real customers, honest answers in the communities where your buyers already ask questions, and coverage you've warranted. It's slower than editing your homepage, and it's where most cited sources actually live.

Measure your citation rate so you know what's working

AI visibility fails quietly — there's no ranking report telling you you've slipped — so track it directly. Run the real buyer questions in your category through ChatGPT, Perplexity, Claude, and Google's AI answers, and record whether you're mentioned, whether you're cited with a link, and which competitors and sources show up instead. Do it on a fixed cadence so you have a trend, not a snapshot, and watch AI-assistant referral traffic as its own channel in your analytics. The point isn't a vanity dashboard; it's to see which moves above line up with citation movement and which don't, so next month's effort goes where the evidence points.

Where I fit

This is the work I do: measure how a company appears in AI search, identify the priorities the evidence supports, and implement the agreed scope. Case Study Zero runs on this site itself, in public, as an unresolved experiment. If you'd rather understand the method than hand it over, everything above is the method. If you'd rather have your baseline measured and the priorities mapped, the contact form is the channel.

Common questions

← All essays