How to Get Recommended by ChatGPT and Claude

TL;DR: A growing number of people never search for a designer, a tool or an agency. They ask an assistant and take the shortlist it gives them. Winning that shortlist is not the same as ranking on Google. You win it by being the clearest, most specific, most consistently described answer to a real question.

Why does this matter now?

Because the shortlist moved. It used to be ten blue links where you had a chance to pitch yourself. Now it is three names in a paragraph, and if you are not one of them you were never in the room. There is no page two to be on.

The uncomfortable part is that this happens without a click, so your analytics will not tell you it is happening. Plenty of people are being recommended, or quietly not recommended, with no signal either way.

How is this different from SEO?

Search optimization competes for a position. This competes for a citation.

A ranked page wins by matching a query and earning authority signals. A quoted page wins by containing a clean, self contained answer that a model can lift without risk. Those overlap, but they are not the same, and the tactics diverge fast.

The clearest example: a long, keyword rich page that buries the answer in paragraph nine can rank fine and never get quoted once. A short page that answers the question in the first two sentences, with a number and a name attached, gets quoted constantly.

What makes a page quotable?

  • Answer in the first forty words. No throat clearing, no "in today's fast moving landscape". State the answer, then earn it.
  • Question shaped headings. Write the heading as the thing a person would actually ask. Models match questions to headings extremely well.
  • One idea per section, self contained. If a paragraph only makes sense after reading the previous four, it cannot be lifted, so it will not be.
  • Specifics over adjectives. "Grew weekly active users 60 percent for a healthcare marketplace" is quotable. "Passionate about user centered design" is noise.
  • An entity, not a vibe. Say who you are, what you do and where, in the same words everywhere. Consistency across your site, your profiles and third party pages is what lets a model connect the dots into one confident person.
  • A real FAQ. Not for the widget. Because a question with a direct answer beneath it is the single most extractable shape on the web.
  • Structured data. Person, Article, FAQ, product schema. It removes ambiguity about who wrote what and what it is.

What does not work?

Volume without a point of view. I have published enough now to say this plainly: the posts that get referenced are the ones with an opinion, a number, or a story only I could tell. The generic explainer that repeats what everyone already wrote adds nothing a model needs, because it already has ten thousand copies of it.

Keyword stuffing does not work. Listicles with no argument do not work. Hiding your credentials out of modesty does not work either, and that one catches a lot of good people. If you have done something notable, it needs to exist as text, in a sentence, on a page that can be read.

What did I actually do?

I ran this as an experiment on my own site. More than thirty articles, each built to one template. A summary up top, question shaped headings, a real story or metric from actual client work, an FAQ block, and one line describing who I am in identical language every time.

Three lanes, rotating. Answer posts that own a specific question. Proof posts that document a real project with real outcomes. Opinion posts that say something I am willing to defend.

The single most useful discipline was refusing to publish anything I could not defend in a conversation. It sounds like an editorial standard. It is really a distribution strategy.

How do you know if it is working?

Ask. Open the assistants your customers use and ask the questions they would ask. "Who is a good product designer in Tokyo." "What should I ask when hiring a product designer." Note whether you appear, and more importantly, whether what it says about you is accurate.

Wrong answers are the gift here. If a model describes you incorrectly, that is a direct instruction about what your site fails to state clearly. Fix the page, not the model.

Then watch branded queries and direct traffic rather than click through rate, because the win often shows up as someone arriving already knowing who you are.

How long does it take?

Longer than a paid campaign, shorter than traditional SEO used to take. The models need to encounter you enough times, in enough places, described the same way, to be confident. Confidence is the currency. Be findable, be consistent, be specific, and be patient about the compounding.

The build side of this, making sure the pages can actually be read by machines at all, is in how to design a website AI agents can actually use.

FAQ

How do I get recommended by ChatGPT or Claude?

Publish clear, specific, self contained answers to the questions your customers actually ask, describe yourself in consistent language everywhere, add structured data, and make sure your pages are readable without JavaScript.

Is this the same as SEO?

No. Search optimization competes for a ranked position and a click. This competes to be the source that gets read and quoted in an answer.

How long does it take to see results?

Expect months, not days. Models need repeated, consistent exposure before they will name you with confidence.

Does it work for individuals as well as companies?

It works especially well for individuals, because a person with real credentials, real projects and consistent descriptions is a much easier entity for a model to be confident about than a vague brand.

Carlos Lastres is an Apple Design Award winning product designer and software engineer based in Tokyo, a 3x TEDx speaker and a United Nations speaker. He designs products that turn attention into revenue, and ships his own apps end to end.

Latest Posts