GEO

ChatGPT SEO: Get Cited, Not Just Ranked

ChatGPT doesn't crawl the web the way Google does, and it doesn't rank your page against nine others.

Schedule a Call

ChatGPT doesn’t crawl the web the way Google does, and it doesn’t rank your page against nine others. When a question is current, specific or commercial, it retrieves from the live web and decides — on its own judgement of which sources are clear about who and what they are — which one to quote. Ranking first in Google is not the same job. This page is about the job ChatGPT actually does: choosing sources it trusts enough to name.

How ChatGPT actually picks a source

ChatGPT answers from training data where it can, and switches to live retrieval when a question needs current information, a specific number, or a commercial decision — “is X worth it”, “how much does X cost”, “X vs Y”. Its retrieval step leans hard on sources that are unambiguous about who and what they describe: a clear organisation entity, a named author, facts stated the same way everywhere they appear. It is less forgiving than Google of a page that hedges or buries the answer under throat-clearing.

Two things matter more here than in classic SEO. First, memory: ChatGPT carries some persistent sense of brands it has seen described consistently and accurately across many sources, and that shapes what it reaches for even before a specific retrieval happens. Second, source discipline: it tends to favour pages that state a commercial fact plainly — your pricing model, what you do and do not do, who you serve — over pages that imply it. Vague marketing copy is exactly the kind of text a retrieval-and-synthesis system struggles to quote with confidence.

What’s included

Entity clarity for ChatGPT specifically. Consistent naming, a coherent organisation entity, and the commercial facts — pricing model, coverage, service boundaries — made explicit in the page copy rather than implied. This is the single highest-leverage change for this engine, more than for Gemini or Claude, because ChatGPT’s retrieval step relies more heavily on unambiguous self-description.

Answer-shaped content blocks. A self-contained 40-60 word answer under a heading that matches the actual question, with supporting detail underneath. Retrieval works on passages, not whole pages.

Source-side coverage for the pages ChatGPT already trusts. Review platforms, industry publications, comparison pages and the sites already ranking in your category. When a brand only talks about itself, ChatGPT has nothing to corroborate against.

ChatGPT-specific prompt tracking. A fixed prompt set covering your buying journey, re-run on a schedule against ChatGPT specifically (not folded into a generic AI-visibility number), recording whether you’re named, which sources back the answer, and how that shifts month to month.

The nine changes that move citation share, applied and prioritised for ChatGPT’s retrieval behaviour: direct answers in the first 60 words, question-shaped H2s, one page per question, named authors with real credentials, first-hand data, current dates and figures, server-rendered HTML, consistent entity facts, and presence on the sources ChatGPT already quotes for your category.

Proof

Official London Theatre reached 85% AI search visibility and the #1 recommended source position in its category, alongside a 67% increase in referral traffic across a five-year partnership. We track ChatGPT as one channel inside that visibility figure, not a separate claim.

Koyo Loans shows the organic precondition for citation: a top-three ranking for “loans” (95% keyword difficulty) and a 73% lift in organic conversions. We haven’t measured what share of ChatGPT answers that ranking feeds — strong organic footing on the fan-out queries is necessary for citation, not sufficient.

Sons is a client we do GEO work for. We’re not publishing prompt-grid or before/after figures for that account until the numbers are sourced and cleared — this page names the relationship without the numbers, per house rule.

How we measure it

Citation share. The percentage of your tracked ChatGPT prompt set where the answer names or links your brand — the headline number.

Source composition. Which domains ChatGPT actually cites alongside or instead of you, so you can see who you’re being corroborated by or against.

Trend, not snapshot. Because ChatGPT is non-deterministic, one run is an anecdote. We report a trend across a fixed prompt set run on a fixed schedule.

You can get a first read for free with our GEO visibility score — no sign-up.

What it costs

There’s no separate ChatGPT invoice. This work is included in every GEO and SEO plan at no extra cost — retainers start from £1,250 a month ex VAT, and splitting search and AI visibility into two line items means paying twice for the same page. Standalone ChatGPT-only GEO work with no retainer attached is scoped to the brief, and either side can end it with 60 days’ notice at any point. Tell us what you need and we’ll price it before we start.

FAQs (8, visible)

How do I get my website cited by ChatGPT?

Answer the specific question directly and early, on a page that’s easy to attribute to a real organisation, and keep the same facts about you consistent on the third-party sources ChatGPT already trusts. In practice: a 40-60 word answer under a question-shaped heading, a named author, first-hand data, and content present in the server-rendered HTML rather than injected by JavaScript.

Is ChatGPT SEO different from Google SEO?

It shares most of its foundations and diverges at retrieval. Crawlability, genuinely useful content and entity clarity serve both. ChatGPT weighs unambiguous self-description more heavily than Google’s ranking algorithm does, and it synthesises an answer from several sources rather than pointing at one page.

Can I pay to appear in a ChatGPT answer?

No. There’s no paid placement in a ChatGPT answer and no submission process. Anyone offering guaranteed inclusion is describing something that doesn’t exist.

Why does my competitor get quoted by ChatGPT when I rank higher in Google?

Usually because they’re described more clearly by sources other than themselves, or because their pages answer the sub-questions behind the prompt while yours answer the head term. Ranking first proves you have the strongest page for a query — it doesn’t prove you have the most quotable passage for a question.

Does schema markup help ChatGPT cite my site?

It supports entity resolution and is cheap to get right, but it isn’t a citation lever on its own — no amount of markup makes a thin page quotable. Check what your pages currently carry with our schema report.

How long before I start appearing in ChatGPT answers?

A page that’s already indexed and just needed a clearer answer can be quoted within weeks. Where the gap is entity clarity or third-party coverage, expect three to six months, because that depends partly on other people’s publishing schedules.

Do I need an llms.txt file for ChatGPT?

We don’t sell one and wouldn’t recommend paying for one. It’s a proposed convention OpenAI hasn’t committed to honouring for ranking or citation purposes, and there’s no evidence today that adding one changes whether you’re cited. If you want it for completeness it takes an afternoon — it isn’t a deliverable we scope work around.

Does ChatGPT remember my brand between conversations?

It can carry a general sense of brands it’s seen described consistently across many sources, which is one more reason entity clarity compounds over time rather than working like a one-off fix.

The Work

GEO campaigns we have run

Latest GEO Insights

Stay ahead with our expert content

GET IN TOUCH

Ready to grow your brand? Get in touch with our team to discuss how we can help you achieve your goals.