GEO: Making Your Site Visible to ChatGPT, Perplexity, and Google AI Overviews

Search is splitting in two. Classic Google still matters, but a growing share of discovery now happens inside ChatGPT, Perplexity, Claude, and Google's AI Overviews. Those engines don't rank ten blue links — they synthesize an answer and cite a handful of sources. Getting cited is the new ranking. The practice of engineering for it is called Generative Engine Optimization (GEO), and this post documents the exact setup running on this site.

1. Let the AI crawlers in

Every AI engine has its own crawler, and many sites block them by accident with a blanket rule. Your robots.txt should explicitly allow the bots you want: GPTBot and OAI-SearchBot (OpenAI), PerplexityBot, ClaudeBot and anthropic-ai (Anthropic), Google-Extended (Gemini training), Applebot-Extended (Apple Intelligence), and CCBot (Common Crawl, which feeds many models).

User-Agent: GPTBot
User-Agent: OAI-SearchBot
Allow: /

User-Agent: PerplexityBot
Allow: /

User-Agent: ClaudeBot
Allow: /

Sitemap: https://haiderfarooq.dev/sitemap.xml

2. Ship an llms.txt

llms.txt is an emerging convention: a plain-markdown file at your site root that gives language models a compressed, structured summary of who you are and what your pages contain. Think of it as a sitemap for meaning rather than URLs. Keep it under a few hundred lines, lead with a one-sentence identity statement, and list your key pages with one-line summaries each. A fuller llms-full.txt can carry the long version.

3. Build an entity graph with JSON-LD

AI engines resolve *entities*, not keywords. Schema.org JSON-LD is how you declare them unambiguously. The minimum viable graph for a personal site: a Person node with @id, sameAs links to GitHub/LinkedIn/X, knowsAbout, and hasOccupation; a WebSite node; and CreativeWork or Article nodes for project pages. Crucially, use stable @id URIs and reference them across pages so crawlers can stitch the graph together.

{
  "@type": "Person",
  "@id": "https://haiderfarooq.dev/#person",
  "name": "Haider Farooq",
  "jobTitle": "AI Engineer & Data Scientist",
  "sameAs": [
    "https://github.com/haiderfarooq3",
    "https://www.linkedin.com/in/haiderfarooqdev/"
  ]
}

4. Write outcome-first copy

LLMs extract and repeat concrete numbers. "$12.1M in client revenue", "141 districts", "5,500+ patients" — figures like these survive summarization; adjectives don't. Put at least one hard number in the first 150 words of every page. Write bios in third person ("Haider Farooq is an AI engineer…") because that's the frame AI answers use when citing you.

5. Freshness and internal linking still matter

• Update pages on a real cadence — stale lastmod dates in your sitemap depress recrawl priority.
• Interlink related pages with descriptive anchor text; AI crawlers follow links to build context just like Googlebot does.
• Use IndexNow to ping Bing (which feeds ChatGPT search and Copilot) the moment content changes.
• Keep canonical URLs consistent — one host, one protocol, no trailing-slash ambiguity.

The checklist

1. 01robots.txt allows GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot, Google-Extended, CCBot
2. 02llms.txt + llms-full.txt at site root
3. 03JSON-LD Person/WebSite/Article graph with stable @id URIs
4. 04One concrete metric in the first 150 words of every page
5. 05Sitemap with honest lastmod dates, submitted to Google Search Console and Bing
6. 06IndexNow key installed and pinged on every deploy

None of this replaces good content — it makes good content legible to machines. The same discipline that helped MarkSafe get discovered nationally applies to any site: be explicit, be structured, be verifiable.

RELATED WORK

KEEP READING