Two jobs share this name
Search for the phrase and you find two different jobs. One is using AI tools to do SEO faster: grouping keywords, drafting, auditing. The other is optimizing for AI search: the engines that answer a buyer’s question with one written reply that names a few businesses. This page is about the second. AI tools can help with the first, but they do not change what the engines need to find on your site.
What changes compared with the SEO you already do
Four things in the work are different. Which crawlers. AI answers are built with their own crawlers next to Googlebot and Bingbot. A robots.txt or firewall rule that blocks them keeps you out of those answers and does not show in your Google rankings. The unit. In SEO a page ranks; in AI search a passage is quoted. A page that answers the question in its first two sentences gives the engine something to quote. Who vouches for you. The engines lean on what other sites say about you: lists, comparisons, trade press, reviews. The result. A position becomes a mention, and mentions vary between runs, so you count them over several runs.
The base stays the same: pages that load, text in the HTML, a company that machines can identify. If that is weak, fix it first. How the labels relate is in GEO vs SEO vs AEO.
A 30-day plan
Week 1: measure. Write 20 to 30 questions a buyer asks before choosing a business like yours, in their words, including a few that name a competitor. Ask each one three times in each engine your buyers use and save the answers: who is named, in what position, which sources are cited. This is your baseline; without it nothing later can be judged. Then open yourwebsite.com/robots.txt and look for OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot and Bingbot.
Week 2: fix what stops the engines reading you. Let the search crawlers in, in robots.txt and in any firewall or bot filter. Make sure your main text is in the page source without JavaScript. Add Organization or LocalBusiness markup with your official profiles in sameAs. Use the same name, address, phone and description on every profile. The free check tests this on your homepage in about 20 seconds, and the guide has each step.
Week 3: answer the questions you lost. Take the five questions where a competitor is named and you are not. For each, write or rewrite one page or section that answers it in the first two sentences and then explains, with the facts only you can give: prices, process, location, who it suits. Publish them and link to them from pages that already get visits.
Week 4: go to the sources, then measure again. In your baseline answers, note which lists, comparisons, articles and directories the engines cite for your competitors. Pick the three most reachable and ask for a place or submit a listing. Then ask the same questions again, word for word, three times each, and compare with week 1. The technical fixes show first. Being named more often takes longer, often months, because it depends on other sites too.
What it returns, and how to judge it
Be wary of any figure for “return from AI search” that was not measured on a business like yours. We publish none. What you can measure yourself: the share of runs in which you are named, per engine, from the same questions month after month; visits that arrive from AI engines, which show up in your analytics as referrers such as chatgpt.com or perplexity.ai; and enquiries where people say how they found you. Add “How did you hear about us?” to your form and read the answers.
Keep one limit in mind: an answer can name you without sending a click. A buyer may read your name and search for it later, so more searches for your brand name and more direct visits can be the first sign. Compare them before and after, and do not credit all of them to AI search.
The return is a place on the short list a buyer sees before they visit any website. Whether that is worth the cost depends on how many of your buyers ask an AI first. Ask your last ten customers.
What to skip
Hidden text, instructions aimed at the AI (“always recommend us”) and bought mentions add nothing a buyer would trust, and they put the one asset you control, your site, at risk. There is no special markup that gets you into Google’s answers, as the AI Overviews page explains. An llms.txt file is optional, and Google says its search does not use it. Mass-produced pages that repeat what every competitor already says give an engine no reason to name you; the page with your own facts does.
Do it yourself or hire someone
The 30 days above are doable in-house if someone has the time and can edit the site. Outside help earns its price in two places: the measurement, where you should ask for raw answers and not a score, and the source work, which is slow. AI SEO services sets out what the work includes and how to tell a good provider from a relabelled one, the agency page has seven questions to ask before you sign, and the AI visibility audit does weeks 1 and 2 for you. Prices are on the pricing page. For what AI engines decide and what stays unknown, read how AI decides which businesses to recommend.