How to Rank in ChatGPT (2026): The Practical AEO Playbook

More and more buying journeys now start with a question typed into ChatGPT, not a search box. When someone asks “what's the best CRM for a hospice agency” or “how do I replace HubSpot,” the model answers in one synthesized paragraph and cites a few sources. This is the AEO game, and unlike classic SEO you can win it on a young domain. Here are the seven moves that actually get you cited.

01What “ranking” means in ChatGPT#

There is no page of ten blue links inside ChatGPT. “Ranking ” means two things: being paraphrased or quoted inside the answer the user reads, and being one of the cited sources ChatGPT Search links beside it. Both are won by being the clearest, most quotable, most reachable answer to a question — not by accumulating the largest backlink profile.

That distinction is the whole opportunity. The link graph that gatekeeps Google rankings took incumbents a decade to build. Answer engines lean far harder on structure and clarity, which a small team can ship this quarter.

021. Let the AI crawlers in#

You cannot be cited from a page the engine cannot read. OpenAI crawls with GPTBot (indexing) and OAI-SearchBot (ChatGPT Search). Many sites still block these by default — the 2026 equivalent of blocking Googlebot in 2005. Open them explicitly in robots.txt, and while you're there allow the rest of the answer-engine fleet: ClaudeBot, PerplexityBot, Google-Extended, and Applebot-Extended.

032. Ship FAQ schema on your key pages#

FAQ schema (schema.org /FAQPage in JSON-LD) is the single highest-leverage move available. It explicitly links a question to its answer, which is exactly the shape an answer engine wants to lift. Add five to eight self-contained Q&A pairs to your three highest-intent pages, written as direct answers a model could quote verbatim.

043. Publish llms.txt and llms-full.txt#

llms.txt hands the engines a curated, plain-text map of your most important pages, and llms-full.txt holds the full canonical text. It removes ambiguity — instead of reconstructing your site from rendered HTML, the model reads the claims you want cited. It's a single static file with outsized return. The full how-to is in our llms.txt guide.

054. Write question-shaped content#

Answer engines match questions, not keywords. Lead a section with the literal question a buyer asks, then answer it in the first sentence — before the context, caveats, and color. The inverted-pyramid style journalists use is the format models reward, because the quotable claim sits right at the top.

065. Build authorship and authority signals#

Models weight E-E-A-T — experience, expertise, authoritativeness, trust — when deciding whose answer to surface. Put a real author with a structured Person schema on every post, cite primary sources, and keep an honest About page. Named, credentialed authorship crosses over cleanly from SEO into AEO.

076. Keep it fresh#

Dated, regularly updated content is cited more than static pages because answer engines try not to quote stale facts. A visible publish-and-updated date, a changelog, and a steady cadence are structural assets now, not vanity. Revisit your cornerstone pages on a schedule and bump them when something genuinely changes.

087. Measure citation lift#

What gets measured gets improved. Run 20–40 category-relevant prompts across ChatGPT, Claude, and Perplexity each month and track three numbers: citation rate (how often you appear), share of voice (your slice versus competitors), and answer accuracy (is the model's snapshot of you current and correct). Trend matters more than any single month.