How to Rank in ChatGPT, Perplexity and Google AI Overviews

More and more questions get answered inside ChatGPT, Perplexity, Claude and Google's AI Overviews without a click to a search result. You can still be the source those answers cite. To rank in ChatGPT and the other AI assistants, you need the same foundations as SEO plus a few new ones: the right crawlers allowed, content that answers questions directly, and clear signals about who you are. This guide covers each step in order of impact.

Where answers come from	How it finds pages	What helps you get cited
ChatGPT search	OpenAI's OAI-SearchBot plus search partner data; ChatGPT-User when a user asks it to open a page	Crawl access, clear answers, strong brand presence across the web
Perplexity	PerplexityBot and its own index; Perplexity-User for live fetches	Crawl access, fresh and well-sourced content
Claude	Claude-SearchBot and Claude-User	Crawl access, clear structure
Google AI Overviews	Google's normal search index	Being indexed and eligible for snippets; normal SEO quality

Step 1: Let AI search crawlers in

You cannot rank in ChatGPT if OpenAI's search crawler is blocked. This is the most common and most fixable problem. Many sites added blanket rules in 2023 to stop AI training and accidentally blocked AI search at the same time. The crawlers are separate, so you can choose:

• Search and user bots (allow these if you want to be cited): OAI-SearchBot and ChatGPT-User (OpenAI), Claude-SearchBot and Claude-User (Anthropic), PerplexityBot and Perplexity-User.
• Training bots (blocking is a legitimate choice and does not affect citations): GPTBot, ClaudeBot, Google-Extended, Applebot-Extended, CCBot and others.

Check your robots.txt group by group; our guide to fixing robots.txt for AI crawlers has copy-paste rules. Also check your CDN or firewall: bot protection that challenges every non-browser request blocks AI crawlers even when robots.txt allows them.

Step 2: Get indexed and fast

AI assistants lean on search indexes. Google AI Overviews only use pages in Google's index that are eligible to show a snippet, and ChatGPT search draws on search partner data as well as its own crawling. So everything in classic technical SEO still applies: pages must be crawlable, indexable, canonical, fast and rendered without depending on fragile client-side JavaScript. Work through the technical SEO audit checklist first; if Google cannot index a page, you will struggle to rank in ChatGPT with it too.

Step 3: Write answer-first content

AI answers are assembled from passages, not whole pages. Pages that get cited tend to make passages easy to lift:

• Answer the question in the first two sentences under a heading phrased the way people ask it ("How long does X take?"), then add detail.
• Use specific facts: numbers, dates, named steps, prices, versions. Vague copy gets paraphrased from someone else.
• Keep one idea per section with descriptive H2 and H3 headings.
• Add FAQ sections and comparison tables, which map neatly onto the questions people ask assistants.
• Cite your sources with outbound links. Well-sourced pages read as more trustworthy to people and to models.

Step 4: Make it clear who you are

When an assistant decides which source to trust, consistent entity signals help. Use the same brand name everywhere, add Organization structured data with your logo and social profiles, show real authors with bios and credentials on expert content, and keep an up-to-date About page. Inconsistent names ("Acme", "Acme Inc", "AcmeHQ") make it harder for models to connect mentions of you across the web.

Step 5: Add structured data

Structured data does not guarantee citations, but it removes ambiguity about what a page is: an article with an author and date, a product with a price, a how-to with steps, an FAQ. Our guide to structured data for AI search covers which types matter, and the schema checker validates your markup.

Step 6: Earn mentions beyond your own site

To rank in ChatGPT for "best X for Y" questions, you need to be mentioned in the places models learn from and retrieve from: reviews, comparison articles, community discussions, documentation and news. This is classic digital PR and community work. Being the brand that independent sources mention is one of the strongest signals, and it is the slowest to build, so start early.

Step 7: Keep content fresh

Assistants favor current information for anything time-sensitive. Update key pages when facts change, show a visible "last updated" date, and keep it honest.

How to appear in AI Overviews

Google AI Overviews are built from Google's own index, so the path is closer to classic SEO than with ChatGPT. To appear in AI Overviews: make sure the page is indexed and not blocked from snippets (a nosnippet directive or max-snippet:0 keeps it out), answer the query directly near the top of the page, and cover the follow-up questions people usually ask next. Google has said no special markup or file is required. Blocking Google-Extended does not remove you from AI Overviews; it only controls use of your content for Gemini training and grounding.

What about llms.txt?

llms.txt is a proposed file that gives models a curated map of your site. It is cheap to add and useful for developer documentation, but no major assistant has said it uses llms.txt to decide citations, and Google does not use it. Read what is llms.txt for the details, and do steps 1 to 5 first.

How to measure whether you rank in ChatGPT

• Referral traffic: look for chatgpt.com, perplexity.ai and similar referrers in your analytics.
• Server logs: confirm OAI-SearchBot, PerplexityBot and Claude-SearchBot actually fetch your important pages.
• Prompt tracking: ask the assistants the questions your customers ask, in a logged-out session, and record whether and how you are cited. Repeat monthly.

How Scanverra helps you rank in ChatGPT

The Scanverra SEO Optimizer includes a generative engine optimization (GEO) section alongside classic SEO checks. It tests your robots.txt against each AI crawler separately and tells you if search and user bots are blocked (which keeps you out of AI answers) versus training bots only (a choice that does not affect citations). It flags pages with no direct-answer block near the top, FAQ content that is not structured as questions and answers, inconsistent brand names, missing author information, no outbound citations, and content that has not been updated recently, and it checks for llms.txt as information only. It cannot tell you how often an assistant actually cites you; use referral data and prompt tracking for that. For a broader view of how ready a site is for AI, see the AI readiness assessment.