AI assistants increasingly answer questions with links to sources. If you want your pages cited, they need to be reachable, readable, and easy to attribute. This checklist covers the checks you can run yourself, roughly in the order you should do them. None of them guarantees a citation. They remove common blockers and give you a way to measure what happens.
1. Check crawler access in robots.txt
Open https://yourdomain.com/robots.txt and look for rules that block AI crawlers. OpenAI uses separate crawlers: GPTBot (training), OAI-SearchBot (search in ChatGPT), and ChatGPT-User (fetches when a user asks ChatGPT to open a page). Perplexity uses PerplexityBot for indexing and Perplexity-User for user-requested fetches. Crawler names change, so confirm them against each vendor's current documentation. A blanket Disallow: / under User-agent: * blocks everything. Blocking GPTBot does not block OAI-SearchBot, so review each agent you care about.
2. Run a server-rendered HTML check
Not every fetcher runs JavaScript reliably. Run curl -s https://yourdomain.com/your-page and confirm that the title, main heading, and key paragraphs appear in the raw HTML. If the page is an empty shell that a script fills in later, move the important content into the server response or prerender it.
3. Add an llms.txt file
llms.txt is a proposed convention: a Markdown file at /llms.txt that gives language models a short map of your site. It starts with a one-line summary, then links to your most important pages with brief descriptions. It is not a formal standard, and not every AI system reads it. It takes little time to write, so treat it as a helpful index rather than a ranking lever. Keep the list to your core pages.
4. Check your sitemap
Make sure /sitemap.xml exists, lists only canonical URLs that return a 200 status, and includes accurate lastmod dates. Submit it in Google Search Console and Bing Webmaster Tools. Some AI search tools draw on search indexes, so checking Bing is worth the few minutes.
5. Add structured data
Add JSON-LD to the page head that describes what the page is. For articles, use Article with headline, author, datePublished, and dateModified. For the site, use Organization with name, url, logo, and sameAs links to your official profiles. Validate the markup with Google's Rich Results Test or the schema.org validator. Schema does not guarantee citations, but it removes ambiguity about who wrote the page and when it was updated.
6. Write answer-first headings
AI systems often lift a passage that directly answers a question. Make each H2 a plain question or task, and put the answer in the first one or two sentences beneath it. Add detail after that. Use "Does robots.txt block ChatGPT?" rather than "A few thoughts on crawlers." Keep paragraphs short and specific, with concrete names, numbers, and dates.
7. Publish a clear about and author page
Models and readers both need to know who is behind a page. Create an About page that says what your site does, who runs it, and how to contact you. Give articles a byline that links to an author page with a real role, relevant past work, and links to profiles you control. Keep claims verifiable and avoid vague superlatives.
8. Use consistent brand naming
Pick one official name and use it everywhere: the site title, page titles, the name field in your schema, social profiles, and the About page. If your brand appears under three spellings, a model may treat them as three different entities. Match capitalization and domain across the web so your pages and mentions point to the same thing.
9. Test real prompts in ChatGPT and Perplexity
Write 10 to 20 questions your ideal reader would actually ask, such as "how do I check if my site blocks AI crawlers?" Run each one in ChatGPT with search turned on, and in Perplexity. Record whether your domain is cited, whether a competitor is, and what the answer says about your topic. Repeat each prompt a few times on different days, because answers vary. Keep a simple spreadsheet with the date, prompt, tool, whether you were cited, and the URL cited. One run tells you little. A trend over several weeks tells you something.
10. Track referrals from chat.openai.com and perplexity.ai in analytics
In your analytics tool, look at referral traffic from AI sources. ChatGPT clicks have shown up with referrers such as chat.openai.com and chatgpt.com, so check both. Perplexity usually appears as perplexity.ai. Create a channel group or segment for these sources so you can see which landing pages they reach and whether those visitors convert. Many AI clicks arrive with no referrer at all, so this number undercounts. Compare it with the citations you logged in step 9.
Wrap-up
Work through the list once, fix the blockers, and then re-run your prompt tests in a month. If you would rather start with an automated pass over some of these checks, there is a free check at citescore.vercel.app/free-check. The manual steps above work without it.
Top comments (1)
We need to output just a comment text, short, one or two sentences, possibly fragment. Should start with specific reaction or question about the video. Could ask about how to detect citation from Perplexia or something. Also include mention of steps. Should be casual, lower case start. No quotes, no markdown. No URLs. No double hyphens. No em-dash. Just plain text. Let's craft: "i tried step 4 on my blog and the link still didn't show up, any idea why the AI