← All notes

Explainer

SEO when there is no click: being the source AI answers with

Answer engines synthesise instead of linking. What that rewards instead of keywords — and what quietly stopped working.

In short

Answer engine optimisation (AEO) is the practice of structuring content so AI assistants — ChatGPT, Perplexity, Google AI Overviews — can extract and cite you directly. Unlike traditional SEO, which competes for a ranked position in a list of links, AEO competes to be the sentence quoted inside a synthesised answer. The tactics that matter most are direct answer-first summaries, consistent entity signals like structured data, and genuinely fast pages.

The click is going away. Referral traffic from Google fell sharply through 2025 as AI Overviews expanded, and the majority of searches now end without anyone visiting a page. If your only goal is a blue link, you are optimising for a shrinking slice of the pie.

Ranking is no longer the whole game

Traditional SEO competes for a ranked position. Answer engine optimisation competes to be quoted — to be the source the model reads and repeats. That is a different goal, and it rewards different structure.

What actually gets quoted

Answer first, prose second

The single highest-leverage change: open every post with two or three sentences that answer its own title outright, before any introduction. A model synthesising an answer needs a self-contained claim it can lift. Burying the answer in paragraph four means it is usually summarised as something vaguer than you wrote.

  • Answer the question the title asks, in the first fifty words.
  • Use real numbers and dates. Vague claims get dropped; specific ones get repeated.
  • Question-shaped headings. They match how people actually query.
  • Keep paragraphs short. Long blocks are expensive to summarise accurately.

Be findable as an entity, not just a page

Structured data is how a machine confirms that the same person is the same person across pages. Schema is not magic — one large study found adding JSON-LD produced no statistically meaningful jump in AI citations. But consistent entity signals, matching author details and identifiers everywhere, reduce the chance you get summarised incorrectly or attributed to someone else.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "BlogPosting",
  "headline": "...",
  "author": { "@type": "Person", "name": "Ojo Gold" },
  "datePublished": "2026-10-11"
}
</script>

Speed is still the entry fee

Core Web Vitals have not changed: LCP under 2.5s, INP under 200ms, CLS under 0.1, judged at the 75th percentile of real visits. INP is now the most commonly failed metric across the web, and it is the one people feel — it measures how long a page takes to respond after you click something.

There is no solid evidence that passing these directly earns you an AI citation. They are a quality floor, not a shortcut. But a static, prerendered page that responds instantly is easier to crawl, cheaper to fetch, and better for the human who does arrive.

The honest summary

The sites that get cited are the ones that say something specific, say it early, are unambiguous about who wrote it, and load instantly. That is not a trick. It is the same discipline that made good pages good, minus the assumption that a visitor is coming.

More Keep reading

Practice · 2 min

Building a website with AI in 2026: the workflow that actually holds up

Coding agents are good enough to build real sites now. The skill is not prompting — it is supervising a loop, and knowing where to stop trusting.

Field note · 2 min

Static sites, edge functions, and the secret that almost shipped

A `.gitignore` is not a security model. Here is what actually keeps an API key off a public website, and the mistake that taught me.