How Utah Businesses Show Up in ChatGPT, Perplexity, and Google AI Overviews

AI assistants name businesses whose websites answer a question directly, in plain text, from a source the assistant can verify. That is the whole mechanism. When ChatGPT, Perplexity, or Google's AI Overviews recommend a Utah plumber or med spa, they are quoting pages that stated something clearly enough to be quoted — and most business websites are written in a way that makes that impossible.

This is being called Answer Engine Optimization, or AEO. Most of it is unglamorous, and much of it overlaps with things good SEO already required.

How is this different from normal SEO?

Traditional search returns a list and the user chooses. Your job was to be on the list, high up. An AI assistant returns an answer, sometimes naming two or three businesses, sometimes one. There is no page two.

Three practical consequences:

Being extractable matters as much as ranking. A page can rank well and still never be quoted, because nothing on it is stated in a form that can be lifted out cleanly.

Corroboration matters more. Assistants prefer claims they can verify in more than one place. A business consistently described across its own site, its Google Business Profile, and independent listings is a safer thing to cite.

Volume matters less; specificity matters more. A thin page that answers one question precisely can be cited where a long page that circles the topic is not.

Why do most business websites fail?

Because they are written as brochures. Consider a typical services page:

"With years of experience serving the Wasatch Front, our dedicated team is committed to delivering exceptional results tailored to your unique needs."

There is no extractable fact in that sentence. No service, no city, no price, no differentiator. An assistant asked "who does X in Utah" cannot use it, because there is nothing to quote.

Compare: "We install and repair tankless water heaters in Salt Lake County. Most replacements are completed in one day and start around $2,800." Every clause is quotable and verifiable.

This is the single highest-leverage change most sites can make, and it costs nothing but writing.

What actually makes a page citable?

Answer the question in the first sentence

Put the answer first, then explain. Journalists call it burying the lede; here it is fatal. If your pricing page opens with three paragraphs about your philosophy before a number appears, the number will not be found.

Use real questions as headings

"How much does a website cost in Utah?" as an H2 with a direct answer beneath it is trivially extractable. "Our Approach to Value" is not. Write headings the way customers actually phrase questions.

Keep facts in plain text

Prices, hours, service areas, and phone numbers must be selectable text in the HTML — not baked into an image, not injected by JavaScript after load. Crawlers that do not execute JavaScript see nothing, and a surprising number of important ones do not.

Add structured data

Schema markup tells machines what things are rather than making them guess: that you are a local business, these are your hours, that is a review with a rating and an author, this is a question and that is its answer. It is not a ranking trick; it is the difference between a machine inferring your details and reading them.

The ones that matter most for a local business are LocalBusiness, FAQPage, and Review. This is part of what our technical SEO work covers.

Be consistent everywhere

Name, address, phone, hours. Identical on your site, your Google Business Profile, and every directory you appear in. Contradictions make you a riskier citation, and assistants resolve risk by picking someone else.

What is llms.txt and do you need one?

It is a plain text file at the root of your domain — yoursite.com/llms.txt — that describes your business and links to your most important pages in a format built for machines rather than browsers. Think of it as a sitemap written in prose.

Honest assessment: it is an emerging convention, not a standard, and support is inconsistent. It will not make you appear in ChatGPT on its own. It costs an hour, it cannot hurt, and very few competitors have one. We publish one at wasatchweb.net/llms.txt if you want to see the format.

Do the fundamentals first. A perfect llms.txt on a site with no extractable content achieves nothing.

Should you block AI crawlers?

There is a real distinction worth understanding. Some crawlers gather training data. Others fetch pages live to answer a question a user just asked — and that second group is how you get cited and how customers reach you.

Blocking the retrieval crawlers to protect content from training is a common and usually self-defeating choice for a local business: you are opting out of being recommended in order to protect marketing copy you published deliberately. If you have strong feelings about training, robots.txt lets you separate the two rather than blocking everything.

The more common accidental version of this problem is a site that blocks or breaks crawlers without meaning to — an aggressive firewall rule, a broken robots.txt, or thousands of 404s eating the crawl budget before real pages are reached.

What should a Utah business do first?

1. Rewrite your top five pages answer-first. Question headings, direct answers, specifics over adjectives. Biggest lever, no budget required.
2. Put your facts in text. Prices, service areas, hours, phone. Out of images, out of JavaScript.
3. Add LocalBusiness, FAQPage and Review schema.
4. Fix your Google Business Profile and make every listing match your site exactly.
5. Publish content that answers real questions. The questions customers ask on the phone are the questions they ask assistants. Each one is a page.
6. Check crawlers can reach you — robots.txt, no firewall rule blocking legitimate bots, no wall of 404s.

Does your website need to be fast for this?

Yes, and for a reason people underestimate. Retrieval crawlers operate on a budget. They fetch a limited number of pages per visit, and a slow site means fewer pages fetched and a higher chance the fetch fails outright.

Two related failures are worth checking specifically.

Content that only exists after JavaScript runs. If your prices, service list, or reviews are injected by a script, a crawler that does not execute JavaScript sees an empty shell. Load your own page with JavaScript disabled and see what survives. Anything that vanishes is invisible to a meaningful share of crawlers.

Crawl budget spent on nothing. Old URLs that now 404, retired landing pages, staging directories left online. Every request spent on a missing file is a request not spent on a page you want quoted. On sites we audit this is routinely the largest single problem, and it is invisible until someone reads the server logs.

Both are fixable in an afternoon, and both sit under ordinary maintenance and technical build quality rather than anything exotic.

How do you know if it is working?

Measurement here is genuinely worse than in traditional SEO, and anyone claiming precise attribution is overselling.

What you can do: ask assistants the questions your customers ask and see who gets named — repeat it monthly. Watch your server logs for AI crawler traffic and check they are getting successful responses rather than errors. And keep asking new customers how they found you, because "ChatGPT recommended you" is currently the most reliable signal available.

Most of this is compounding, unglamorous work — which is also why the businesses that do it tend to keep the advantage for a while. If you would rather it were handled, technical SEO and monthly content are both part of our $299/month plan, or tell us what you are working on and we will point you at the parts that matter most for your situation.