Nukebook
Market & Site Intelligence
Platform page

Claude SEO: what is actually known.

A new term with no incumbent answer. Here is what is documented, what is measurable, and what is guesswork wearing a lanyard.
If you searched "claude seo"

What we measured about it: 390 times a month of demand at difficulty 0 of 100: a citation surface with no incumbent answer worth displacing.

Run on your market, a read shows: which AI surfaces cite you today and which are winnable, plus your real search demand under every phrasing, which AI answers in your market are open, partial or closed (with the evidence), what your competitors run in the public ad archives, and your break-even numbers computed from your own economics.

Book the 20-minute callor read the measured pieces first →

What is Claude SEO?

Claude SEO is the practice of making your content findable and citable by Anthropic's Claude when it searches the web to answer questions. Claude fetches pages at request time and cites the sources it uses, so the practical work is classic citability: crawlable server-rendered pages, self-contained passages that answer one question completely, and figures with sources and dates. Anthropic documents its crawlers publicly, so access can be verified in your server logs rather than guessed at. (The term is searched 390 times a month in the US at difficulty 0 of 100; Nukebook, 2026-08-28.)

"Claude seo" is searched about 390 times a month at difficulty 0 of 100 (top bid $18.24). The sibling phrasings are real too: "llm seo" 880 searches a month and "chatgpt seo" 390 searches a month. measured

What is documented versus what is vibes

Documented: Anthropic publishes its crawler user agents, so you can check your own logs for whether Claude can and does fetch your pages, and robots.txt controls apply. Measurable: whether Claude's answers on your buyers' questions cite you, checkable by asking and recording, repeatedly, over time. Guesswork: anyone selling a ranking factor list for Claude. Model behavior shifts; across our own two-pull measurements of AI answers, 7 changed between pulls in a single week.

The practical checklist

Serve real HTML, not a JS shell. Answer one question per passage, completely, near the top. Carry figures with sources and dates: models under citation pressure prefer checkable claims. And measure: ask the model your buyers' questions monthly and record whom it cites. That record is the only Claude SEO metric that is not made up.

Method: query metrics from DataForSEO (2026-08-28); crawler facts limited to what Anthropic documents publicly; answer-volatility figures from our two-pull August 2026 runs. Figure registry: /figures/claude-seo.figures.json.
Want this measured for your market?

One read: your demand, your winnable AI answers, every competitor's ads, and your site's catch rate. Price and scope on a short call, plainly.

Book a call or see what a read contains →