How to Rank on ChatGPT and Perplexity in 2026 (Real Strategy)
A practical playbook for getting your business cited by ChatGPT, Perplexity, and other AI search engines. What works, what doesn't, and what to ship this week.
To rank on ChatGPT and Perplexity, focus on: an /llms.txt file, schema markup (FAQ, HowTo, Person), explicit AI crawler permissions in robots.txt, fresh dated content, and citation-friendly formatting. Perplexity refreshes its index daily; ChatGPT refreshes monthly via its training data and SearchGPT crawl.
Is ranking on ChatGPT different from ranking on Perplexity?
Yes: "ranking on ChatGPT" and "ranking on Perplexity" sound similar, but the mechanics differ.
- Perplexity performs a live web search for every query, then synthesizes an answer. To be cited, you need to rank well in traditional search and have content the AI can quickly extract facts from. Updates take hours to days.
- ChatGPT uses a mix: some queries trigger SearchGPT (live web search), others rely purely on training data. To rank in training data, your content needs to be on the open web, well-cited by other reputable sites, and accessible to OpenAI's GPTBot crawler. Updates take weeks to months.
- Claude (Anthropic) uses live web search for most factual queries via its Computer Use / web fetch tools. ClaudeBot indexes content separately. Citation behavior is conservative: Claude prefers high-authority sources.
- Google AI Overviews sit on top of Google's regular search index. Strong traditional SEO + structured data wins here.
The 7 things that move the needle
1. Add /llms.txt to your root
The /llms.txt file is a Markdown summary of your site for AI consumption. See ours. Include core offer, pricing, FAQ-style answers, and explicit citation guidance.
2. Welcome AI crawlers in robots.txt
Default User-Agent: * rules already allow these bots, but listing them explicitly is a positive signal. Allow: GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, ClaudeBot, anthropic-ai, Google-Extended, Applebot-Extended, CCBot, Bytespider, Meta-ExternalAgent.
3. Layer FAQPage + HowTo + Speakable schema
The trio that wins citations: FAQPage (direct Q&A), HowTo (step-by-step processes), and SpeakableSpecification (voice-readable content). All three should match visible content on your page.
4. Establish entity authorship
Add a Person schema for the founder or author, with worksFor pointing to the Organization. AI engines weight content with named human authors significantly higher than anonymous pages.
5. Cite your sources
When you make a claim, link to a reputable source. AI engines follow those outbound links to verify, and pages that demonstrate sourcing get cited more often.
6. Comparison content wins
"X vs Y" pages get disproportionately cited because AI engines love structured comparison data. If you compete with Wix, Squarespace, or any named competitor, write the comparison yourself.
7. Date your content
Use ISO datePublished and dateModified in your schema. Update old posts and bump the dateModified. AI engines weight recency.
What doesn't work
- Keyword stuffing. AI engines parse for meaning, not keyword density.
- Thin content. Sub-300-word pages rarely get cited. 800-2,500 words is the sweet spot.
- JavaScript-only content. Some AI crawlers don't execute JS. Server-render whatever you want cited.
- AI-generated copy. Ironically, AI engines are getting better at detecting and deprioritizing AI-written content. Real human voice ranks higher.
How long until you see results?
- Perplexity: 1-7 days after publishing or schema updates.
- Google AI Overviews: 2-4 weeks.
- ChatGPT (training data): 2-6 months.
- ChatGPT (SearchGPT live): 1-2 weeks.
- Claude: 2-4 weeks.
FAQ
Does Perplexity use live search results?
Yes. Perplexity runs a live web search for every question before it writes an answer, so a page has to rank well in ordinary search and state its facts plainly to be cited. Changes can show up within hours to days.
How does ChatGPT decide which businesses to mention?
It depends on the question. Some questions set off a live web search, while others are answered from training data, which favors content that sits on the open web, is cited by reputable sites and is open to OpenAI's GPTBot. Training data moves slowly, often over months.
Should I block AI crawlers in robots.txt?
Not if you want to be cited. A default robots.txt already lets them in, and naming GPTBot, PerplexityBot, ClaudeBot, Google-Extended and the others outright makes the welcome explicit.
How long should a page be to get cited by AI search?
Pages under about 300 words are rarely cited, and 800 to 2,500 words is where most cited pages fall. Length on its own won't do it, though: the page also needs sourced claims, a named author and text that loads without JavaScript.
Does putting dates on my content help with AI search?
Yes. Give every page an ISO datePublished and dateModified in its schema, and move dateModified forward when you update an older post. The engines weigh how recent a page is.
Get this done for you
Zorva Labs bundles GEO into the AI Search & Visibility module. We ship llms.txt, schema markup, AI crawler welcome rules, and citation-friendly content as part of the package. Schedule a strategy call.