What the published numbers actually say
Pew Research Center analyzed the browsing data of 900 US adults in March 2025. Fifty-eight percent of them ran at least one Google search that returned an AI-generated summary. When a summary appeared, they clicked a traditional search result in 8 percent of visits, against 15 percent when no summary appeared. They clicked a link inside the summary itself in 1 percent of visits.
Read that carefully, because it cuts both ways. Fewer clicks is bad news if your business model is web traffic. It is close to irrelevant if your business model is a phone call, because the customer still saw the answer, and the answer still named somebody.
Why local is different from publishing
Most of the alarm about AI search comes from publishers, whose whole revenue depends on the click. A local service business is not in that position. Nobody clicks through to a plumber to read an article. They read the recommendation, then they call, or they search the business name, or they open the map. The recommendation is the conversion event, and it happens off your analytics.
That is the actual risk for a local owner, and it is not a traffic risk. It is that you can be steadily losing the recommendation without a single number on any dashboard moving.
So is it worth attention yet
Proportionately. It is not yet where most local demand comes from, and anyone telling you to move your whole budget is overselling. It is growing, it is cheap to measure, and the work that improves it (profile accuracy, service pages, review text) is the same work that improves ordinary local search. The realistic position is to spend an hour a month on measurement and let the fixes ride along with what you already do.
Detecting AI-sourced customers with what you already have
Four detection methods, none of which need a new tool. Set all four up in an afternoon and you will have a signal within a month.
| Method | How to set it up | What it tells you, and what it misses |
|---|---|---|
| Referral hostnames in analytics | In your analytics referral report, look for chatgpt.com, perplexity.ai, gemini.google.com, claude.ai, and copilot.microsoft.com. | Counts only people who clicked. Real but a severe undercount. Treat any number above zero as a floor, not a total. |
| The intake question | Add one line to your call script: how did you hear about us, and were you searching or asking an assistant? | The only method that catches the call-not-click path, which is the majority. Needs staff discipline to be worth anything. |
| Branded search volume | Watch Search Console impressions and clicks for your business name specifically. | AI recommendations produce name searches. A rising branded curve with flat generic traffic is the classic fingerprint. |
| Direct mention tracking | Ask the assistants your category's buyer questions on a fixed schedule and record whether you are named. | The leading indicator. It moves weeks before the phone does, and it is the only one that tells you which questions you are losing. |
Run all four. Each is blind to something the others catch, and the first three are lagging measures of the fourth.
You are looking for the term
Zero-click search
The pattern where a customer gets what they needed without visiting anyone's website is called zero-click search, and it is why AI visibility is measured in mentions rather than sessions. If your analytics look flat while your market is clearly changing, this is the phrase that explains the gap.
- My analytics show no AI referrals at all. Does that mean nothing is happening?
- Not necessarily. Referral tracking only catches people who clicked a link, and the published click rates are very low. The call-intake question and branded search volume will show you what analytics cannot.
- Should I stop investing in Google and move to AI?
- No. Google is still where most local demand is, and the work that wins AI answers is largely the same work that wins local search. Treat this as an addition to measurement, not a replacement for a channel.
- 01Do people click on links in Google AI summaries?Pew Research Center
- 02From Googlebot to GPTBot: who is crawling your site in 2025Cloudflare
- 03Analytics dimensions and metricsGoogle Analytics Help
- 04AI features and your websiteGoogle Search Central
Every link above was opened and confirmed reachable on September 1, 2026.
The automated version
See what ChatGPT, Claude, Gemini and Perplexity say about your business
The free scan runs the buyer questions for your trade and city across all four assistants, shows who gets named instead of you, and lists the sources those answers came from. No signup, about thirty seconds.
Run my free scan