What this is, and why you should care
Your customers are asking AI products questions about your category, and the answers they get are deciding who they call, buy from, trust and ignore. That is the whole story. The search box used to be ten blue links, and now it is a paragraph written by a model that has quietly chosen which businesses to put in it. If your name is not in that paragraph, you have a problem the old SEO scoreboard cannot see and will not warn you about.
The good news is that being in the paragraph is mostly the same craft that has always made a business findable on the open web, with a handful of extra moves on top. The bad news is that the rules for those extra moves are different on every engine, they shift while you are reading about them, and a lot of the advice currently being sold as gospel is somebody guessing in a confident voice. This guide is the calm version. We tell you what to do, we explain why, and where the field is genuinely uncertain we say so.
Known & Cited is a GEO-led PR consultancy. We tell you, plainly and with evidence, how ChatGPT, Google AI Overviews, Perplexity, Claude and Gemini are talking about your business right now, using our AI Visibility Strategy (AVS). We then help you do something about it.
The 12 things below are standard things that everyone can do to help nudge things in the right direction:
// SECTION 0212 things to do, in plain English
These are roughly ordered by value for effort, and any business that takes its presence online seriously should do the first five before anything else on the list.
01Let the right AI crawlers in.
Every AI product sends one or more bots to fetch your pages, and if those bots are blocked in your robots.txt file your content quite literally cannot be cited because the model has never seen it. Allow Googlebot, Google-Extended, GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, Perplexity-User, ClaudeBot, Claude-Web, CCBot and MetaExternalAgent at the minimum, and check today rather than tomorrow because this is the single most common own-goal we find in the wild.
02Use one brand name, consistently, everywhere.
AI products rely heavily on what is called named-entity consistency, and if your homepage says Acme Corp while your contact page says Acme Limited and your LinkedIn says Acme Group, you are quietly handing the model three different entities and asking it to pick a favourite. Pick one canonical name, put it on every page and meta tag, lock it into your Organisation schema, and tidy your third-party listings so they all sing the same note.
03Put a real person's name on every page.
Author bylines with role, credentials and contact information matter more than people think, because the models are trained to lean on what Google calls E-E-A-T (experience, expertise, authoritativeness, trustworthiness), particularly in regulated subjects like health, finance and law. A page written by "the team" is a page that struggles to be cited, full stop.
04Date your content and show a visible last-updated stamp.
AI products prefer fresh and time-stamped sources because their retrieval systems are built around the assumption that newer is usually truer, and if your article is undated the model has no way of knowing whether it was written last week or in 2019. Publish a date, show a last-updated date when you revise, and you have given the retriever a signal it actively rewards.
05Cite your primary sources inline.
If you make a claim, link to the regulator, the original study or the official guidance behind it, because AI products reward pages that show their working in the same way a sceptical editor would. A page that cites primary sources is far more likely to be picked up as a primary source itself, and that compounds over time.
06Write content with a first-hand point of view, not the commodity version.
"Seven Tips for First-Time Homebuyers" is content the model can produce from training data without ever leaving the building, whereas "Why We Waived the Inspection and Saved Money: A Look Inside the Sewer Line" is something the model genuinely needs your page for. Aim for the second every time, even when it feels narrower than your marketing instincts want.
07Structure each page for a human reader who is in a hurry.
Use clear headings, short paragraphs, scannable sections and lead with the answer before the detail, because AI products extract passages rather than whole pages and a clean passage is much easier to lift than a wall of text. There is no ideal page length, despite what you may have read elsewhere. Make pages as long as the subject genuinely needs and no longer.
08Build internal links that reflect how your topics actually relate.
The hub-and-spoke pattern works because it tells crawlers and retrievers the shape of your expertise on a subject. A primary topic page sits at the centre, deeper pages on the sub-topics radiate out, and links travel back and forth so the relationships are obvious rather than implied.
09Earn coverage in places AI products already cite.
Tier-one trade press, established publishers and reputable charities or institutions are wildly over-represented in AI citation data, and a single genuine mention in the Guardian or the FT is worth more than a thousand syndicated reposts of your own content. Real PR is earned and editorially honest. Citation farming is a waste of money and the models are getting better at spotting it.
10Be present where your buyers genuinely talk to each other.
Reddit, Stack Overflow, niche forums and trade community boards are retrieved heavily by Perplexity and Google AI Overviews in particular, and an honest, helpful presence in the right communities does more for your visibility than another blog post pushed to your own mailing list.
11Use structured data as Search hygiene, not as a miracle.
Schema markup for Organisation, Person, Article, FAQ and Product helps Google show rich results and helps several AI products parse your pages cleanly, but it is not required for AI features and it is not a silver bullet. Do it once, do it properly, and then stop tinkering.
12Repeat your category positioning across your pages in your own words.
If you sell AI brand intelligence, the phrase "AI brand intelligence" should appear naturally on your homepage, your about page, your services page and across several blog posts, because repetition reinforces the link between your named entity and your category in the model's understanding. Repeated and intentional.
// SECTION 03Generic guide, tailored work
This guide is generic by design and we have written it that way on purpose, because the twelve moves above are the floor that almost every business should stand on before they do anything cleverer. The ceiling is different. You cannot beat a tailored approach built around exactly where you want to show up, across the engines, the audiences and the buying journeys that actually matter to your business, and that is the work we do at Known & Cited. The guide is the map. A tailored AI Visibility Strategy engagement is the route.
// SECTION 04What each engine seems to favour at the moment
The notes below are necessarily imperfect because the products move quickly and rarely document themselves in detail, so we treat them as working hypotheses and revise them when our measurement tells us otherwise.
ChatGPT
ChatGPT combines older training data with live retrieval and tends to favour authoritative third-party publishers and Wikipedia-grade sources, which means brand mentions in long-form journalism punch above their weight.
Google AI Overviews and AI Mode
Google AI Overviews and AI Mode are powered by Gemini but grounded in the normal Search index and Knowledge Graph, so freshness, structured local and ecommerce data, and classical SEO fundamentals all matter, and of all the engines this is the one closest to "just do SEO properly".
Perplexity
Perplexity favours fresh primary sources, well-structured listings and B2B-grade authority, and it cites Reddit and trade press more readily than the others.
Claude
Claude leans heavily on long-form, well-edited legacy journalism and reference material, is less inclined to cite forum content than Perplexity, and has a conservative citation tone.
Gemini
Gemini in its standalone app behaves slightly differently to Gemini inside Search, and we treat it and AI Overviews as one bucket while flagging where they diverge.
// SECTION 05What we do not yet know
We do not know exactly how each product weighs its sources because the vendors publish hints rather than facts. We do not know whether agentic retrieval, where AI agents browse and act on your site directly, becomes the dominant access pattern over the next twelve months. We do not know how long any of these specifics will hold, and six months from now half this list could need a footnote.
Our position is consistent and it has not moved since we started. Measure first, then act on what the measurement actually surfaces, and treat anyone selling certainty in AI visibility as someone selling certainty rather than measuring it.
Focus. Measure. Plan. Deliver. Repeat.
If you want to know how you are being cited right now across ChatGPT, Google AIO, Perplexity, Claude and Gemini, our free AVS exec brief is the place to start, and the tailored AVS engagements take it from there. Just get in touch.
Free AVS Exec Brief
See how the five major AI products are talking about your business right now. A real one-day measurement using the same methodology as the full AVS.
Get the Free AVS Exec Brief