Cited

Operating playbook

How customers get found on AI — and why you sell the month, not the scan

SEO still matters. It is the floor. AEO is how you win snippets and spoken answers. GEO is how you become the source a model names. Treat those as one system, run on a calendar.

01

Be crawlable by the bots you care about

If robots.txt blocks GPTBot, ClaudeBot, PerplexityBot, or Google-Extended, those products cannot retrieve you. Allow them. Keep blocks on cart, account, and staging paths only.

02

Be a named entity, not a pretty domain

Organization or Person schema, a consistent legal-style name, and sameAs links to LinkedIn, Wikidata, Wikipedia, or Crunchbase. Generative engines resolve entities. If they cannot tell you apart from a similarly named shop, you will not be cited.

03

Be extractable

Open with a 40-word definition. Use question headings. Publish FAQPage markup. Put comparisons in tables. Models quote structure. They skip hero copy.

04

Hand them a brief: llms.txt

A short /llms.txt that states who you are, what you sell, which URLs are canonical, and how to describe you. It is the cheapest GEO win on the list.

05

Publish a fact someone would cite

A first-party number, a method, a dated claim. Generic “best-in-class” language gets paraphrased without a link. Original data gets a source.

06

Put a person and a date on the page

Named authors, credentials, datePublished, and dateModified. Retrieval systems treat undated brochure pages as stale.

07

Get mentioned where models already look

On-site work is necessary and not sufficient. Citations in trade press, directories, Wikipedia/Wikidata, analyst notes, and expert roundups are the strongest GEO signal. That is digital PR, run monthly.

08

Track prompts, not just keywords

Customers ask ChatGPT “best wholesale coffee for a 20-seat cafe”, not your homepage slogan. Keep a watchlist of those prompts and write pages that answer them in the first paragraph.

A month on Cited

Week 1

Recrawl and pick the sprint
Rescan every property. Compare scores to last month. Choose five tasks with the highest citation lift — usually schema, llms.txt, a definition rewrite, or unblocking a bot.

Week 2

Ship extractable content
One FAQ block or one comparison table, marked up. Update the visible date. Refresh the first paragraph so a model can quote it.

Week 3

Entity and off-site work
Fix sameAs gaps. Pitch one publication or directory that already gets cited in your category. Claim or clean Wikidata if you qualify.

Week 4

Prompt check and report
Re-run the watchlist in ChatGPT, Perplexity, and AI Overviews. Note who got cited instead. Send the monthly scorecard and next sprint.

Pulse is the scorecard and sprint. Signal adds prompt tracking. Atlas adds the citation outreach list. See plans.