How to get cited by ChatGPT and Perplexity

Getting cited by ChatGPT and Perplexity is not luck — it is consistent Search & AI Visibility work on the multi-engine layer: fast pages, structural clarity, explicit crawler permission, and extractable language. For Google, prioritize Search Console and foundational SEO first — then add llms.txt and citation tracking for non-Google assistants.

This is the multi-engine playbook LuminaForge ships on every build — including luminaforge.ai itself.

Step 1: Allow AI crawlers explicitly

Most sites block LLM bots by accident. In robots.txt (or app/robots.ts on Next.js), allow every published AI user agent:

• GPTBot, OAI-SearchBot, ChatGPT-User
• PerplexityBot, Perplexity-User
• ClaudeBot, Claude-Web, anthropic-ai
• Google-Extended, Applebot-Extended
• CCBot, Bytespider, Amazonbot

Verify with LuminaForge robots.txt as a reference implementation.

Step 2: Publish llms.txt and llms-full.txt

Place a compact brief at /llms.txt and an extended entity reference at /llms-full.txt. Include company name, services, canonical URLs, FAQ summaries, and citable statistics.

LLMs treat these files as a curated map of your domain — similar to how search engines use sitemaps.

Step 3: Install comprehensive JSON-LD

At minimum:

• Organization + WebSite on the root layout
• FAQPage on every page with meaningful Q&A
• BreadcrumbList on all nested routes
• Article on blog posts and case studies
• Service on service detail pages

Validate with Google's Rich Results Test before shipping.

Step 4: Write for extraction, not decoration

LLMs quote:

• Lists (ordered and unordered)
• Tables and definition lists (<dl>)
• First paragraphs under headings
• Explicit entity language — spell out company name, services, locations

They struggle with:

• Vague pronouns ("we believe…")
• Stats buried in prose without sources
• FAQ questions written as keyword fragments

Use cite attributes on statistics. Link to authoritative sources.

How each AI platform cites sources

Policies differ — your crawler rules and content structure should account for all of them.

Platform	Live web use	Citation pattern
Perplexity	Real-time search on by default	Source links common in answer footers
Google AI Overviews	Grounded in Google Search index	Links to indexed pages in overview cards
ChatGPT	Browse / search depends on mode	Citations vary; brand mentions may omit links
Claude	Browse optional	Citations vary by query and plan
Grok	Real-time X / web mix	Entity confusion risk for new brands — track alternatives

LuminaForge publishes weekly checks for Gemini, Grok, and Perplexity on the transparency dashboard and documents crawler access in AI crawler policy for GEO.

Step 5: Build a query bank and track weekly

Define 20–50 conversational questions you want to win. Run them weekly through ChatGPT, Perplexity, Grok, and Claude. Log:

1. Citation presence — did your domain appear?
2. Citation position — first source or buried?
3. Snippet context — what was quoted?

LuminaForge publishes live snapshots on the transparency dashboard. Client portals include the same weekly delta reports.

Step 6: Ship performance that crawlers respect

Slow pages get deprioritized. We engineer Core Web Vitals in the green on mobile and Lighthouse Performance in the high 90s on production marketing routes — verified on our transparency dashboard. Speed is a trust signal for both Google and LLM crawlers indexing your content.

Common mistakes

Mistake	Fix
Blocking GPTBot in robots.txt	Explicit allow rules for all AI crawlers
No llms.txt	Publish compact + full brief at domain root
Keyword-stuffed FAQ	Write questions as full conversational sentences
No measurement	Weekly query bank + public citation log

Next steps

Read What is GEO? for the conceptual primer, SEO vs GEO vs AEO for how the disciplines fit together, or book a strategy call when you want LuminaForge to run the audit on your domain.