How to get cited by ChatGPT and Perplexity

When a prospective customer asks ChatGPT or Perplexity who to trust in your field, one of three things happens: the model names you, it names somebody else, or it hedges and names nobody. If you are unhappy with how often the first one happens, read on, because this is a solvable problem. Shaping those answers is the practice of AI search optimisation, and it is less mysterious than it sounds. Models cite the businesses they can find, verify and trust, and there is a great deal you can do to be all three. Mostly it is the PR work that has always mattered, done more deliberately.

Here is how those citations actually get earned, and how to tell whether it is working.

How the models choose who to cite

An AI answer is not a ranking of links, it is a synthesis. The model draws on your footprint across the whole web: the coverage you have earned, the places you are mentioned, how consistently your story appears from one platform to the next. It is, in effect, judging your reputation at scale.

Consistency matters more than most people expect. If the story on your website, your LinkedIn and your press coverage all differ, the model's confidence drops, and confidence is what produces a citation rather than a hedge. If it cannot find enough about you at all, it may fill the gap with something generic, or something subtly wrong, which is worse.

What AI search optimisation actually involves

Start with clarity. These systems reward expertise stated plainly: real answers to the questions your buyers actually ask, in the language they would use, published under a real name. If your positioning only makes sense to people who already know your category, the models cannot work with it, and neither can most customers.

Then presence. Models learn about you from the places they read: respected publications, video, consistent profiles, mentions by other people. Earned coverage carries the most weight, because a third party vouched for you. Video deserves particular attention; a meaningful share of AI recommendations cite YouTube, which surprises founders who treat it as a social channel rather than a search asset.

And beneath all of it, authorship. Named experts, a consistent entity, a story that does not change depending on where it is found. If that sounds familiar, it is the argument I made when I wrote about why authorship is the new SEO, and it applies doubly here.

Be the source, not just the subject

There is a difference between being written about and being quoted, and the second is worth more. Being featured says you exist. Being the expert a journalist calls for comment says you are an authority, and it is repeatable: each quote makes the next one likelier. That is the loop to build.

It is also where a relationship-led approach pays off. I spent my career around the media before founding Adia, and what I learned still holds: journalists need reliable, quotable experts, and they return to the ones who deliver. Become one, in public, and the press and the models both start reaching for your name.

The mistakes that keep you invisible

Some of what passes for AI-readiness is noise. A flood of thin, keyword-stuffed articles does more harm than good, because the models are good at spotting filler and so are people. Republishing the same press release on a dozen wire services adds nothing either; it is one fact wearing twelve coats.

The other mistake is going quiet. A burst of coverage followed by silence reads, to a person, like a business that had news once. To a model it reads like a fading signal. Whatever cadence you can sustain, sustain it.

Here is the monthly routine I would run to stay citation-ready:

• Ask ChatGPT and Perplexity the questions your customers actually ask, and note who gets named and why.
• Publish one clear, named answer to a real question your buyers have.
• Pitch one piece of expert commentary to a publication the models already read and cite.
• Check your story matches everywhere: website, social profiles, recent coverage.
• Log every mention you earn, so the footprint builds under one consistent name.

How to tell if it is working

This is not instant. A footprint takes months to shift, and the answers shift with it. The test is simple and free: ask the same questions every few weeks and watch what changes. Do more names appear? Does yours appear with the right positioning attached? Is the hedging shrinking?

There is a compounding effect at work, too. Early on you will be cited occasionally, alongside competitors. As your coverage thickens and your entity becomes unambiguous, the hedges give way to names, and the names narrow. The only way through it is a steady volume of genuine signal over time.

It is worth the patience. When a model recommends you by name, that is not a vanity metric, it is the new shop window, and it works while you sleep. The businesses being cited today started building their footprint before anyone was talking about any of this.

If you would like an outside view on why the models are not saying your name yet, book a discovery call. We will look at your footprint together and talk about what would move it, with no pressure either way.