Notes — No. 3
How to be the answer when customers ask AI
Ask an AI assistant who should redo a kitchen in Naples and it doesn't hand back ten links — it hands back three names. Nobody on that shortlist bought the spot. They were legible. This note is about what that means, and how to become one of the names.
From ten links to one answer
For two decades, being found meant ranking on a results page. That page is dissolving. Google now answers many searches itself, with AI Overviews stitched together above the links. Millions of people skip the search engine entirely and ask ChatGPT, Perplexity, or Claude directly — the way they'd ask a well-connected neighbor. However the question gets asked, the answer has changed shape: not a directory to browse, but a recommendation to act on. Two or three names, with reasons.
For a local business, that's a narrower door than a results page ever was. Page one of Google had ten spots and a map pack. An AI answer has about three. And the businesses occupying them aren't necessarily the biggest — they're the ones the machine could read, verify, and confidently repeat.
What the machines actually read
An assistant preparing an answer behaves like a careful, skeptical researcher. It reads your website — the actual text, not the design. It cross-checks your Google Business Profile: hours, address, categories, reviews. It looks for corroboration in directories, maps, news mentions, and social presence. And it weighs agreement heavily. When your site, your profile, and the wider web all say the same thing about what you do and where, the machine treats you as established fact. When they contradict — an old address here, different hours there — it treats you as uncertain, and uncertainty is disqualifying in an answer with three slots.
Beneath the visible page sits a second layer these systems lean on: structured data — a machine-readable statement of your business's facts, written in a shared vocabulary called Schema.org, invisible to visitors and unambiguous to machines. A page can succeed with people and fail this layer completely. Text trapped in images, menus that exist only as PDFs, sites that need a human's intuition to navigate — to a machine, these are locked doors.
GEO, defined
Generative engine optimization — GEO — is the practice of making a business easy for AI systems to read, verify, and recommend. It isn't a bag of tricks, and it doesn't replace SEO; assistants read the same web Google indexes, so the disciplines overlap. The difference is the goal. SEO competes for position on a list. GEO works to be quotable — to be the business whose facts are so clear and so consistently corroborated that a machine can state them with confidence, in its own words, to a customer who will never see your homepage.
What this looks like in practice
The work is unglamorous, which is exactly why so few businesses have done it. State the basics in plain text on your own site: what you do, precisely where you do it, when you're open, what a customer can expect to pay or how to find out. Answer, on the page, the questions customers actually ask — the ones you hear on the phone every week. Make your name, address, and phone number identical everywhere they appear, character for character. Put structured data behind every page so the facts are stated to machines directly instead of left to inference. And treat reviews as infrastructure rather than vanity: a steady flow of specific, detailed reviews is one of the few forms of reputation a machine can verify.
None of this requires abandoning beauty. It requires building so the beauty and the facts coexist — a site that moves a person and informs a machine in the same breath.
The honest caveats
Nobody can guarantee placement in an AI answer, and anyone selling guaranteed rankings in ChatGPT is selling weather. These systems change monthly, their reasoning is partly opaque, and measurement is young — referral traffic from assistants is only beginning to show up in analytics in a usable way. What can be said honestly is this: legibility compounds. Every factual, consistent, verifiable signal you put into the world is read by every current system and every future one. The businesses that did this work early for Google owned page one for a decade. The same window is open again, and it is not crowded — least of all in Southwest Florida.
Questions owners ask
What is generative engine optimization (GEO)?
Generative engine optimization is the practice of making a business easy for AI systems — ChatGPT, Claude, Perplexity, Google's AI Overviews — to read, verify, and recommend. Where SEO competes for position on a results page, GEO works to make your business one of the few names an assistant gives when someone asks who to call. It rests on clear content, consistent facts, and structured data.
How is GEO different from SEO?
SEO optimizes for a ranked list of links; GEO optimizes for a synthesized answer. AI systems don't show ten results — they name two or three businesses and explain why. That favors pages that state facts plainly, answer real questions directly, and agree with the rest of the web about your name, address, hours, and services. Good SEO still matters: assistants read the same web Google indexes.
How do AI assistants decide which businesses to recommend?
They read the open web — your website, your Google Business Profile, reviews, directories, news mentions — and look for corroboration. A business whose story is consistent everywhere, whose site plainly states what it does and where, and whose reputation is verifiable gets treated as trustworthy. Contradictions, thin pages, and information locked inside images or scripts make a business hard to verify, so it gets left out.
Do Google reviews affect whether AI recommends my business?
Yes, meaningfully. Reviews are one of the few reputation signals a machine can verify at scale, and assistants lean on them when deciding who to name. Volume, recency, specificity, and your responses all carry weight. A steady flow of detailed reviews that mention specific services does more good than a burst of five-star ratings with no text.
What is structured data, in plain English?
Structured data is a small block of code on your website that states the facts of your business in a format machines read directly — name, location, hours, services, who runs it — using a shared vocabulary called Schema.org. Visitors never see it. Search engines and AI systems do, and it removes the guesswork from understanding your page. It's the difference between being interpreted and being understood.
Can I pay to show up in AI answers?
Mostly no, and be wary of anyone who says otherwise. Sponsored placements are beginning to appear in some AI products, but the recommendations themselves are drawn from what the systems read and verify on the open web. The durable investment is legibility — a clear site, consistent information, real reviews — because that's what every current and future assistant draws on.
How can I check what AI says about my business?
Ask the assistants what your customers would ask — 'best kitchen remodeler in Naples,' 'who should I call for pool service in Bonita Springs' — in ChatGPT, Perplexity, and Google's AI mode. Note whether you appear and whether the details are right. Then read your own website and ask whether a stranger could state your services, area, and hours from it alone. That gap is the work.
I run Saltwind Media, a digital studio in Naples. I build web presences designed to be found and trusted — by people and by the AI systems that increasingly answer on their behalf. That includes reading your business the way a machine does and telling you plainly what it sees. If you'd like that read-through, write me.