Emerging Digital PartnersGet in touch

Article · content

Your answer is on the page. It is just 600 words too late.

A 2026 analysis found 44 percent of ChatGPT citations came from the first third of the cited page, and only around 15 percent of retrieved pages were cited at all. Warming up for four paragraphs before saying anything is not a style choice any more. It is the reason your competitor gets quoted.

By George McKenna3 min read

Open the last article your business published. Count the words before it says anything a customer could act on. In our audits the median is around 300, and 600 is not unusual. Four paragraphs on why the topic matters, a definition everyone already knows, a note about how things have changed, and then, eventually, the answer.

That structure was harmless when a human skimmed to the bit they needed. It is expensive now, because the reader that decides whether you get cited stops scoring long before word 600.

How an engine reads

An AI engine does not read your page. It retrieves it, cuts it into passages, scores each passage against the user's question, and quotes the passage that answers most directly, with the page as the citation. It does this across dozens of candidate pages at once and keeps very few. Research in early 2026 found ChatGPT cited only around 15 percent of the pages it retrieved. Of the citations it did make, 44 percent came from the first third of the page.

Two things follow. The unit of competition is the passage, not the page. And position within the page matters in a way it never did for classic ranking, because the passages near the top are the ones most likely to be a direct answer to the question the page is titled for.

The habits that bury answers

Most of them are inherited from an earlier era of SEO and content marketing:

The keyword intro. Two paragraphs designed to repeat the target phrase before saying anything. Engines understand synonyms; Google's guide says so explicitly. The intro adds nothing and pushes the answer down.

The importance preamble. "In today's competitive market, choosing the right X has never been more important." The reader knows. The engine does not care.

The suspense structure. Building to a conclusion, as taught in essays. Good for essays. Fatal for a page whose job is to answer "how much does X cost".

The label heading. H2s that say "Overview", "Benefits", "Our approach" instead of stating a question or a claim. An engine scoring passages against a question gets no help from "Benefits".

The hedge. "It depends on a number of factors" as the opening line of the pricing section, with the actual range three paragraphs later.

What answer-first looks like

The answer in the first two sentences of the page. Then each section opens with its own conclusion and spends the rest of the section justifying it. Headings are the questions customers ask, in their words. Specifics (numbers, names, places, prices) are in the first sentence under each heading, not buried in the explanation.

Before: "There are many factors that influence how long a website redesign takes, and every project is different. In this article we will explore the key phases and what affects the timeline."

After: "A small business website redesign usually takes six to ten weeks: one to two for discovery, two to three for design, two to four for build, and a week for testing and launch. Content delays are the most common cause of overruns."

Same information. The second version is a citation waiting to happen; the first is a promise to answer later, and the engine does not wait.

This is not chunking

Google's May 2026 guide told site owners they can ignore "chunking": breaking content into fragments on the theory that AI systems need it pre-cut. Google's systems, it says, can find the relevant passage in a multi-topic page. That is true of the other engines too; it is exactly what passage retrieval does.

Answer-first is different. The page stays whole and coherent. What changes is the order inside each part: conclusion, then reasoning. A human skimming headings and first sentences gets the entire argument. So does an engine. Nothing is fragmented; things are simply not delayed.

The uncomfortable part

Answer-first writing commits you. Publishing "six to ten weeks" or "from £3,500" or "we do not work with clients outside Sussex and Kent" means saying it plainly, at the top, where a competitor and a customer can both see it. The preamble habit is partly a way of avoiding that commitment.

The engines have removed the option. Someone in your market is publishing the number at the top of the page, and the assistant is quoting them. The only question is whether the number it quotes is yours.

Frequently asked questions

Where on the page do AI engines take their citations from?

+

Disproportionately from the top. One February 2026 study of ChatGPT citations found 44 percent came from the first third of the page. Engines score passages against the question and lift the one that answers most directly, so the earlier a clear answer appears, the better its odds.

Isn't restructuring pages for AI what Google told us not to do?

+

Google said not to chunk content into fragments or rewrite it for keyword variants. Opening each section of one coherent page with its conclusion is neither. It is what Google's helpful content guidance has asked for all along: organise content so readers can find what they need.

Sources

  1. Search Engine Land: 44% of ChatGPT citations come from the first third of content · searchengineland.com
  2. KIME: ChatGPT citation sources decoded · kime.ai
  3. Google Search Central: Optimizing for generative AI features · developers.google.com

George McKenna

Co-founder, Emerging Digital Partners

George co-founded Emerging Digital Partners in Eastbourne and built the AI Search & SEO Audit you are on now. He spent around twenty years in the UK IT channel before that, most recently running solution sales teams, and now splits his time between building websites and search tooling for EDP clients and taking AI products to regulated industries through the ETT Group, where he is Chairman and CTO. He writes about answer engine optimisation the way he practises it: test it, measure it, fix what is actually broken.

Article15 Sept 2026 · 3 min read

AI cannot quote a homepage that says nothing

'We deliver innovative solutions that empower your business.' An AI engine reads that and learns nothing it can use. The vague-copy problem was tolerable when a human read past it to the contact form. It is fatal when the reader is a machine deciding whether to recommend you.

George McKenna
contenton-pagestrategy
Article15 Sept 2026 · 5 min read

No author, no proof, no chance: the trust gap on small business websites

Most small business websites ask to be trusted without offering a single checkable reason. Anonymous posts, stock photos, no team page, no dates, no credentials, no named clients. A human visitor forgives that. An AI engine deciding whether to recommend you does not, because it has nothing to weigh.

George McKenna
contentstrategy
Article15 Sept 2026 · 4 min read

You are the only one talking about you

Every large citation study in 2026 lands in the same place: AI engines lean on Reddit, Wikipedia, YouTube, LinkedIn, editorial sites and review platforms, and cite brand-owned pages far less than owners expect. A business whose only source about itself is its own website is asking to be taken on.

George McKenna
strategyai-search

← All articles