# Your customer asked an AI instead of Google — and you weren't in the answer

> A growing share of customers now ask ChatGPT or Google's AI mode for a recommendation instead of searching. If a site isn't readable by machines, it never makes those answers.

Canonical page: https://septminor.online/en/blog/ai-search-visibility/
Site overview: https://septminor.online/llms.txt

Published: August 19, 2026 · 6 min read · Author: [Septminor](https://septminor.online/en/about/)

A year ago the path to your business was predictable: someone opened Google, typed "dental clinic near me," looked at the first few results and called one of them. Today a growing share of those people type nothing at all. They ask ChatGPT or Gemini: "recommend a dentist in my area with evening appointments who's good with anxious patients" — and get back two or three specific names with a short explanation of why those. If your name isn't among them, you never learn that person existed. No click, no impression, no line in any report. The customer simply wasn't there.

## This isn't a forecast — it already shows up in the reports

Google built the same conversational mode into its own search, and since August 2026 unusual rows have started appearing in Search Console (Google's free report showing which queries bring people to your site). Not keywords, but fragments of an actual conversation: "yes, go on," "for a small team," "how much does that cost," "how is it different from the first one." Google even sorted these phrases into roughly seven types — clarifying, comparing, asking about price, signalling readiness to buy, and so on.

The interesting part isn't the new rows themselves. It's that an intermediary has appeared between your customer and your website. Before, the only thing standing there was Google's results page, where position decided everything. Now there's also an AI that visits sites on the person's behalf, picks a few of them, and retells what it found in its own words — while the person sees a finished answer and one or two recommendations.

## What a machine sees when it lands on your site

A site built for people and a site a machine can read aren't automatically the same thing. A human lands on a service page, sees a nice banner, infers from context that you work nationwide, opens an accordion and reads the price. An AI agent infers nothing from context: it works with the text that is actually on the page and with claims it can verify. If prices only exist inside an image, terms are explained in a video, and there are no plain answers to the simple questions a customer asks, then from a machine's point of view the site is close to empty. Not bad — empty. There's nothing to retell.

That's why the businesses surfacing in these answers often aren't the biggest players in the market. They're the ones whose site simply states, in ordinary sentences, what they do, who they do it for, what it costs and how they differ from the company next door.

## Why it costs money even though nothing in your reports drops

The uncomfortable part is that this loss doesn't look like a loss. Ads keep running, cost per lead holds steady, inquiries haven't collapsed — no number in any dashboard turns red. A new channel of demand is simply growing past you, and every month a competitor with a well-written site makes it into those answers while you don't, they take customers who never even reached your site. That kind of damage doesn't show up in a weekly report; it shows up in market share a year later.

It's the same blind spot as people who visit your site and leave without converting — covered separately in [a visitor left your site without converting — and vanished from your ads for good](https://septminor.online/en/blog/remarketing-for-business/). The difference is that here the person never reaches the site at all. And to notice where inquiries actually come from and what each one really costs, you need analytics that's properly set up — otherwise new channels stay invisible far longer than they need to; more on that in [what a lead really costs without Google Analytics 4](https://septminor.online/en/blog/ga4-setup/).

## What usually goes wrong

The most common reasons a business doesn't make it into AI answers — and none of them are about budget:

- Prices, terms and timelines exist only inside an image, a PDF or a slide deck. A person can read that; a machine can't.
- Pages are full of promises ("individual approach," "market leaders") and contain no direct answer to what the customer actually asks: what it costs, how long it takes, whether you work in their industry.
- Key information is hidden inside pop-ups and collapsed blocks loaded by script. For most AI crawlers that text doesn't exist on the page at all.
- Company details are scattered and inconsistent: one name in the header, another in the contacts, a service list on the homepage that doesn't match the service pages.
- The site is technically blocked to newer AI crawlers — in hosting settings, at the CDN, or in robots.txt, often without the owner knowing, "on by default for security."
- Nothing said on the site is corroborated anywhere else: no case studies with numbers, no reviews, no mentions on third-party sites. Models are cautious about claims that appear nowhere else, and will name a competitor with external mentions instead.

None of this means replacing advertising with "AI optimisation" — paid channels aren't going anywhere and will stay the main source of leads for a long time. It means there's now one more way customers find a business, and getting into it costs less money than attention to what your site actually says.

Two more pieces dig into this same shift: why [organic traffic is down 40% and it is not a seasonal dip](https://septminor.online/en/blog/organic-traffic-drop-is-not-a-slump/), and why [AI recommends your competitor instead of you](https://septminor.online/en/blog/why-ai-recommends-competitors/) — and it is not about site authority.

## Find out whether AI can see you

I'll look at your site the way an AI agent does: what it can actually read, what's missing before it would name you in an answer, and which of those gaps are quick to close. You get a concrete list of fixes, not general advice.

Have something to say about marketing, AI, or digital entrepreneurship? [Suggest an article for the blog →](https://septminor.online/en/suggest-article/)