Being cited by AI search

There is a lot of confident advice about optimising for AI, and most of it is unfounded. The parts that hold up are unglamorous.

Be reachable first. An assistant cites what its index contains. Before optimising anything, check the boring things: pages that only exist after JavaScript runs, a sitemap declared but never created, and missing pages returning your error page with a 200 status so every wrong URL looks like real content.

Be the best answer to a narrow question. Broad topics are the least winnable for a small site. If a dozen established sources already answer it well you will not be cited. If the honest state of the art is a forum post from three years ago, you can be the best source by the afternoon.

Answer rather than tease. Content designed to make someone click withholds the answer by design, and that is precisely what cannot be cited. Put it in the page, including the parts that make you less necessary.

Be sceptical about measurement. Some tools report prompts you defined and how you scored. Others report genuinely observed citations. Only the second says anything about the world, and confusing them will have you celebrating the day you added prompts to a tracker.

Equally, be sceptical about tactics with a good story and no evidence behind them. Some widely recommended files and formats have no measured effect at all.

The 11 notes in this area

Change detection that fires on every deploy is worse than none

Hashing a rendered page to spot real content changes sounds simple. Modern frameworks rewrite enough of the document on every build that almost every page looks changed.

Your best-ranking pages might not link to what you sell

Guides and explainers often outrank product pages by fifty positions. If they do not link to the thing they describe, that ranking does nothing for you.

How to get cited by AI search

What actually moves AI citations: being crawlable, being specific, being the best answer to a narrow question, and getting new pages indexed fast.

IndexNow: getting new pages indexed in minutes

A simple protocol for telling search engines a URL changed, instead of waiting to be crawled. What it does, what it does not, and how to set it up.

Reading only Google Search Console hides how you are actually found

A second search engine can be a fifth of your impressions, a third of your clicks, and the entire basis of your visibility inside AI assistants. It has its own console.

A keyword tool gives you a hypothesis, not a target

A term with good volume and low difficulty can still be the wrong phrase. Your own impression data is the only thing that confirms buyers use those words.

llms.txt probably is not doing what you think

The file is cheap and harmless, and the measured evidence says AI crawlers are not fetching it and publishing it does not affect citations. Spend the effort elsewhere.

A rate limited crawl will invent problems that do not exist

If your site auditor records a failed fetch and an empty field the same way, every rate limited URL becomes a fabricated defect, and the report looks completely normal.

When you retract a claim, the copy is the easy part

A wording change gets applied to the pages everyone looks at. It survives in the articles, the structured data and the product feed that keeps publishing it to shopping engines.

Your sitemap is advertising pages that do not exist

Two hand-maintained lists, one of published articles and one of sitemap slugs, drift apart. The sitemap keeps pointing search engines at pages nobody ever wrote.

Soft 404s: when your error page returns success

A one line web server misconfiguration that makes every wrong URL on your site look like a real page to a crawler.

Other guides

• Running n8n in production
• Running an AI support desk
• Automating orders, payments and returns
• Knowing where your traffic actually comes from
• Email that actually arrives
• Self-hosting it reliably