How Do You Get Your Business Cited by ChatGPT, Perplexity and Google AI Overviews?
Your customers have started asking ChatGPT, Perplexity and Google who to hire. If those tools can't read your site — or can't find a passage worth quoting — they recommend someone else. That's the whole problem Generative Engine Optimization (GEO) exists to solve.
- AI Overviews cut organic CTR for position-1 results by 58% (Ahrefs, Dec 2025, 300k keywords).
- Brands cited inside AI answers get 35% more organic clicks (Seer Interactive, 2026).
- The fix has four layers: crawler access, quotable content, structured data, real URLs.
What is GEO, and why does it matter now?
Generative Engine Optimization is the practice of making your business easy for AI systems to read, trust and cite. The stakes became measurable in 2025: Ahrefs analyzed 300,000 keywords and found that when an AI Overview appears, organic click-through for the #1 result drops 58% (Ahrefs, 2025).
Sounds grim? Here's the other half. Seer Interactive tracked 3,119 informational queries across 42 companies and found that being cited inside the AI answer produced 35% more organic clicks and 91% more paid clicks than not being cited (Search Engine Land, 2025). Traffic didn't vanish. It moved — from ranked lists to cited sources.
Which AI crawlers do you need to allow?
Every major answer engine sends its own crawler, and each one respects robots.txt. If your robots file — or your firewall, or your CDN's bot protection — blocks them, you are invisible to that engine, no matter how good the content is. Checking this takes two minutes and fixes the most common GEO failure we see.
From our own setup: this is the exact allowlist running on georgeai.dev right now — GPTBot and OAI-SearchBot (OpenAI), ChatGPT-User (live browsing), ClaudeBot and anthropic-ai (Anthropic), PerplexityBot and Perplexity-User, Google-Extended (Gemini), Applebot-Extended, Bytespider and CCBot (Common Crawl, which feeds many model training sets). Each gets an explicit Allow: / so the policy is unambiguous.
Why be explicit instead of relying on the wildcard? Because bot-protection layers and future robots edits tend to break the implicit case silently. An explicit allowlist documents intent.
What is llms.txt, and do you need one?
llms.txt is a plain-markdown file at your site root that gives language models a curated summary: who you are, what you sell, which pages matter. Think of it as a sitemap written for machines that read prose instead of XML. Vendor adoption is still uneven — treat it as cheap insurance, not a guarantee.
Ours describes each service in one line with a link to its page. One page of markdown, ten minutes of work, and any engine that does read it gets exactly the summary we'd want quoted.
How should you structure content so AI can quote it?
AI answers are assembled from passages, not pages. The unit of citation is a 40–60 word block that makes sense with zero surrounding context: one claim, one number, one named source. If your key facts are spread across three paragraphs of storytelling, there's nothing extractable to quote.
Three rules cover most of it. Open every section by answering the heading's question directly, in the first sentence. Keep one idea per paragraph. And give every important claim a number and a source an engine can verify — vague authority ("industry leaders agree") gets skipped.
Ask yourself: if a stranger read only this paragraph, would it still be true, complete and attributable? That's the passage an AI will lift.
Does schema markup still matter for GEO?
Yes — arguably more than for classic SEO, because schema is how engines resolve entities. A JSON-LD graph that links your Organization, founder, WebSite and each Service with stable @id references tells an AI "these facts belong to this one business" instead of leaving it to guess from prose.
From our own setup: georgeai.dev ships a connected graph — ProfessionalService → founder → Person, a WebSite node, and a Service + BreadcrumbList node on every service page, all cross-referenced by @id. Same email, same phone, same name in every node. Consistency is the signal.
Why do real URLs beat a single-page app?
Answer engines cite URLs. If your whole site is one JavaScript-rendered page, every service, case study and answer competes for a single citation slot — and crawlers that don't execute JavaScript see almost nothing. Giving each service its own static, crawlable URL multiplies your citation surface.
We learned this on our own site: the portfolio started as a pure SPA, and the single biggest GEO improvement we made was generating a real page per service — fifteen indexable URLs where there had been one. Each service now has its own page, title, description and schema.
How do you measure whether it's working?
Ask the engines. Monthly, put your money questions to ChatGPT (with browsing), Perplexity and Google's AI mode — "best AI automation agency for small businesses", "who builds AI phone agents" — and log whether you're cited, linked or ignored. Pair that with server logs: GPTBot and PerplexityBot hits confirm you're at least being read. Organic CTR on AIO-present queries bottomed at 1.3% in December 2025 and recovered to 2.4% by February 2026 (Seer Interactive, 2026) — this channel is still moving, so measure monthly, not yearly.
Want this done for your site?
Our SEO / AIO / GEO service implements everything above — crawler access, llms.txt, schema graph, citable pages — with a fixed scope and a before/after visibility report. Or start with the configurator for a bundled quote.
Frequently asked questions
Is GEO different from SEO?
They overlap heavily. Solid SEO (crawlability, structure, authority) remains the foundation. GEO adds AI-specific work: allowing AI crawlers, writing self-contained quotable passages, llms.txt, and entity-rich schema. Ahrefs found AI Overviews cut position-1 CTR by 58% in 2025 — ranking alone no longer guarantees traffic.
Should I block AI crawlers to protect my content?
If you sell to people who ask AI tools for recommendations, blocking GPTBot or PerplexityBot removes you from those answers entirely. Seer Interactive measured 35% more organic clicks for brands cited inside AI Overviews. For most service businesses, visibility is worth far more than content protection.
Does llms.txt actually do anything yet?
Adoption by AI vendors is still uneven, and no platform guarantees it reads the file. It costs one page of markdown, though, and gives answer engines a clean, curated summary of who you are. We treat it like early schema markup in 2012: cheap insurance that may compound.
How long until AI engines cite a small site?
Expect weeks to months, not days. AI answers lean on crawlable, well-structured pages plus third-party mentions. In our experience, real crawlable URLs per service and consistent entity data (same name, email, schema everywhere) move the needle before anything else does.