Getting found in AI search: how AEO actually works

For the past twenty-five years, "getting found online" meant getting found on Google. SEO was the discipline. Page one of the search results was the goal.

The first thing changing is where people look. ChatGPT, Claude, Perplexity, Google's own AI Overviews, and emerging AI shopping assistants are increasingly the first stop for the questions that used to start a Google search. A buyer who used to type "best AI consultant for small business in Western New York" into Google is now asking the same question of ChatGPT.

The second thing changing is what wins. The factors that make a website rank in classic Google search are not the same factors that make an AI engine cite it as a source.

AEO — Answer Engine Optimization — is the practice of making your business visible and citeable in this new layer. This post covers what we actually do for clients at Genesee AI to help them show up.

What AI engines are doing under the hood

When a user asks ChatGPT or Claude a question, the model does one of two things:

1. Answers from training data. What it learned during its last training run, frozen as of some cutoff date.
2. Searches the web live. Most modern AI assistants now run web searches in real time when the question is recent or specific.

For a brand to be found in case (1), it needs to have been part of the public web in a way the model's training data captured. For case (2), it needs to be findable by the search layer the AI uses (often Bing, sometimes Google, sometimes a custom index).

In both cases, the wins come from having clear, factual, structured content about who you are and what you do.

What helps you get found

A few patterns hold consistently across the AI engines:

• A clear, factual home page. First paragraph that states what the business is, who it serves, what it does, where it is. No marketing fluff at the top — the AI is looking for facts and will skim past adjectives.
• A FAQ section. Structured Q&A is the format AI engines extract most reliably. Real questions, real answers, brand name in both.
• Schema markup. Organization, ProfessionalService, FAQPage, Article — structured data that tells the engine exactly what your business is. Most websites do not have it. The ones that do get cited more.
• A /llms.txt file. A new (and rapidly adopted) convention: a plain-Markdown file at the root of your domain that summarizes the site for LLM crawlers. The AI engines that respect it index your content faster and more accurately.
• Mentions and citations on third-party sites. Press, podcasts, partner sites, directories, reviews — these reinforce to the AI that your brand is a real entity worth knowing about.
• Content depth on your specific service areas. A blog post that answers "what does an AI sales coaching deployment look like" by name, in plain English, gives the AI something to cite when a user asks the same question.

What hurts you

The same things that hurt in classic SEO, mostly:

• Thin or duplicate content. AI engines weigh originality.
• No clear topic focus. A site that is about everything is about nothing.
• Missing or wrong factual basics. If your contact page lists a city you no longer operate in, the AI will tell people the wrong thing.
• Blocking crawlers. Some sites accidentally block GPTBot, ClaudeBot, or PerplexityBot in their robots.txt. Check yours.

What we typically build at Genesee AI

A standard AEO engagement for a small or mid-sized business looks like:

1. A content audit. What is on the site, what is structured well, what is invisible to AI crawlers. A factual gap analysis.
2. Schema markup. Full structured data for the organization, services, FAQ, and any individual content pages.
3. An llms.txt file. Clean Markdown summary of the business at the root of the site.
4. A targeted content plan. Six to twelve posts that answer the specific questions buyers ask AI engines about your category. Each post is a candidate citation.
5. An off-site signal plan. Press, podcast appearances, directory listings, partner mentions. The work that reinforces to the AI engines that you are a real entity.
6. A measurement loop. Quarterly checks across ChatGPT, Claude, Perplexity, and Google AI Overviews — does the brand come up for the queries it should? What is being said? Where is it being cited from?

What about classic SEO

Still matters. AI search and classic search overlap heavily — the engines often pull from the same underlying web index. Most of the work that helps you rank in Google also helps you show up in AI engines.

The shift is that you no longer need to rank #1 on a search results page. You need to be one of the three to five sources an AI engine pulls from when a buyer asks a question in your category. Lower bar to be cited, higher bar to be the only source.

A realistic timeline

AEO is a months-not-weeks discipline. The infrastructure changes (schema, llms.txt, content) can ship in days. The actual lift in AI-engine visibility usually shows up over the next two to three quarters, as crawlers re-index your site and content gets cited in subsequent AI training runs and real-time searches.

If you want a baseline read on how AI engines currently describe your business, book a free consultation and we will pull the current picture across the major engines.