When someone asks an AI assistant for "the best tool for X", the answer is assembled from pages the assistant can fetch and from what the model already knows about you. You can't buy a spot in that answer, but you can remove the reasons you're left out. This checklist is ordered by impact.
1. Let the AI search bots in
Each assistant uses its own crawlers. If your robots.txt or CDN blocks them, you can't be cited no matter how good your page is. The ones that fetch pages to answer users are:
- OAI-SearchBot and ChatGPT-User (ChatGPT search and browsing)
- PerplexityBot and Perplexity-User (Perplexity)
- Claude-SearchBot and Claude-User (Claude)
- Googlebot (Google Search and AI Overviews) and Bingbot (Bing and Copilot)
Check your robots.txt and any 'block AI bots' switch in Cloudflare or your host. Blocking training crawlers like GPTBot or ClaudeBot is a separate choice: it doesn't stop search citations, but future models may know less about you.
2. Make the page readable without JavaScript
Most AI crawlers read raw HTML and don't run JavaScript. If your homepage is a client-rendered app shell, they see an empty page. Server-render or pre-render your key pages and say plainly what the product does, who it's for and what it costs.
3. Answer the questions people actually ask
People prompt AI with questions: "X vs Y", "is X good for Z", "how much does X cost". Pages with a clear FAQ and honest comparison sections map directly onto those prompts and are easy to quote.
4. Tell machines who you are
Add schema.org JSON-LD: an Organization with your name, logo and sameAs links to your social profiles, plus a SoftwareApplication or Product with a description and offers. This removes ambiguity about which company or product a page is about.
5. Publish llms.txt
llms.txt is a short markdown file at your site root that points language models and agents to your most important pages. It's cheap to add and increasingly read by agents. See our guide to llms.txt for the format.
6. Be mentioned elsewhere
Assistants weigh what other sites say about you: directories, reviews, comparison articles, GitHub, Reddit and community threads. Consistent descriptions across these sources make it easier for a model to recommend you confidently.
Check where you stand
RankRoot's free scan checks points 1 to 5 on your homepage in about ten seconds and generates the llms.txt, robots.txt and JSON-LD for you.