How to optimize for AI search: an 8-step checklist

Step 1: Check where you stand today

Before changing anything, record a baseline.

1. Write down 20 to 50 questions a buyer might ask an AI tool about your category, such as "best accounting software for contractors" or "who does same-day furnace repair in Denver".
2. Ask each one in ChatGPT, Perplexity, Gemini and Google (watching for an AI Overview).
3. For each answer, note whether you are mentioned, whether you are linked, and which competitors and sources appear instead.

Then run our free AI visibility audit. It checks AI crawler access, schema, llms.txt and entity signals, which are the technical items in the steps below. It is the same audit our AI SEO agency uses to scope client work.

Step 2: Let AI search crawlers into robots.txt

Each AI company publishes its own crawlers, and each robots.txt rule works on its own. The ones that matter most for being found:

OpenAI recommends allowing OAI-SearchBot if you want to appear in ChatGPT search answers, and Perplexity recommends allowing PerplexityBot. If you do not want your content used for model training, you can disallow GPTBot while keeping OAI-SearchBot allowed. A simple setup that allows search but opts out of OpenAI training:

User-agent: OAI-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: GPTBot
Disallow: /

Also check your CDN, firewall and bot protection. A robots.txt allow does nothing if your security layer blocks the request first. OpenAI says robots.txt changes take about 24 hours to register on its side, and Perplexity says up to 24 hours.

Step 3: Make sure you are indexed in Google and Bing

Google says a page must be indexed and eligible for a snippet to appear as a supporting link in AI Overviews or AI Mode. Check Google Search Console for pages excluded from the index and fix the causes, such as noindex tags, canonical mistakes, thin duplicates or crawl errors.

Do the same in Bing Webmaster Tools. Other AI assistants are widely reported to draw on Bing's index, so it is worth the few minutes it takes to verify your site and submit a sitemap.

Step 4: Add structured data that matches the page

Google describes structured data as "explicit clues about the meaning of a page" and recommends JSON-LD. Start with:

• Organization (or LocalBusiness) on your homepage, with your name, logo, URL, contact details and sameAs links to your official profiles.
• Article or BlogPosting on content pages, with author and date.
• Product, Service or SoftwareApplication where they fit.
• BreadcrumbList for site structure.

Keep the markup consistent with the visible text. Google lists that as part of its advice for AI features. Note that Google limited FAQ rich results to well-known government and health sites in 2023, so add FAQ markup for clarity, not in the hope of rich results.

Step 5: Publish an llms.txt file

llms.txt is a proposed standard, published by Jeremy Howard in September 2024, for a Markdown file at /llms.txt that gives AI agents a short overview of your site and links to your most useful pages. The format is an H1 with your site name, a one-line summary in a blockquote, and H2 sections listing links.

Be realistic about it. Google says you do not need new machine-readable files or AI text files to appear in its AI features, and the main AI search providers have not documented using llms.txt as a ranking input. It is cheap to add and helps agents that do read it, so treat it as low-effort housekeeping, not a growth lever. Our llms.txt generator builds one for you.

Step 6: Make your entity data consistent

AI models form a view of your business from many sources at once. If your website says one thing, your Google Business Profile another and an old directory something else, the model has less reason to describe you confidently.

• Use the same business name, address, phone, service list and one-line description everywhere.
• Link your official profiles from your site, and from your Organization schema with sameAs.
• Update or remove outdated listings.
• Keep your About page factual: who you are, what you do, where, and for whom.

Step 7: Rewrite key pages answer-first

AI tools quote passages, not whole pages. Make those passages easy to lift.

• Open every important page with a 2-4 sentence answer to its main question.
• Use question headings and answer directly underneath each one.
• Put steps in numbered lists and comparisons in tables.
• Add specific facts: prices, ranges, timeframes, locations, specs. Vague claims give a model nothing to cite.
• Cover the follow-up questions too. Google says its AI features may run several related searches across subtopics, a process it calls query fan-out, so pages that cover a topic's subtopics have more chances to be used.

The research paper that coined the term GEO found that adding quotations and statistics were among the more effective methods it tested, with gains varying by subject. Our generative engine optimization guide covers this in more depth.

Step 8: Earn citations and track prompts

When someone asks an AI tool for a recommendation, the answer often reflects comparison articles, review sites, directories and forum threads. Look at the sources cited in your Step 1 checks, then work to get listed or mentioned on them: request inclusion in relevant roundups, collect reviews on the platforms your buyers use, and contribute useful answers where your audience asks questions.

Re-run your prompt set every month. Track mention rate, citation rate and how you are described, and compare against the baseline from Step 1. Our AI visibility tools roundup lists software that automates this.

If you would rather hand the whole checklist off, our AI search optimization services do exactly this for a flat monthly fee: month one fixes what the audit found, then we monitor up to 50 prompts a month across ChatGPT, Perplexity, Gemini and Google AI Overviews and keep improving the results. For platform-specific detail, see how to rank in ChatGPT.