Does zero-click search cost you customers?
Published on
It costs clicks, that is measured; customers, that depends on the company. Pew Research Center observed the Google searches of 900 U.S. adults in March 2025. When an AI summary was displayed, they clicked a traditional link in 8% of visits, versus 15% without a summary, and a link in the summary in only 1%. Those who lived off visits from people seeking information lose the most. A company that sells something loses mainly when the answer cites others and not it: the customer does not click, but has still chosen.
For what type of question? This is a question about the commercial effect of answers that are displayed without anyone clicking: Google's AI summaries and the assistants' answers. The available measurements count clicks and visits, in the United States; none directly counts lost customers.
What Pew measured
Pew Research Center analyzed the browsing of 900 U.S. adults from a panel who accept a tracking app, from March 1 to 31, 2025: 68,879 Google searches. 18% of them displayed an AI summary. When a summary appeared, users clicked a traditional result in 8% of visits; without a summary, in 15%, almost twice as often.
The links placed in the summary itself were clicked in 1% of visits. And users more often ended their browsing after a page with a summary: 26% of cases, versus 16% without a summary. Pew notes that news publishers attribute their traffic losses to these summaries. Its figures date from March 2025, and Google's AI answers may have changed since.
Fewer clicks does not mean fewer customers
A click is not a customer. For a site that lives off reading, an informational article, a guide, a page that earns money through advertising, the zero-click answer replaces the visit: the loss is direct. For a company that sells a product or a service, the customer ends up contacting someone. The question becomes: whom did the AI recommend to them?
The visits that remain are worth more. Adobe observes that in September 2026, among U.S. retailers, visits coming from an AI led to a purchase 90% more often than others. Someone who clicks after an answer has already done their sorting. Fewer visits, but visits closer to a purchase: for a cited company, the loss of clicks is not necessarily a loss of sales.
Where the loss really happens
The loss happens in the answer itself. When a customer asks which supplier to choose, the AI names a few companies. Those that are named keep a chance. The others receive neither a click nor a visit, and see nothing in their statistics. G2 measured this among B2B software buyers in March 2026: 69% chose a different vendor from the one they had planned on, because it appeared in the AI's recommendation.
The same survey shows the other side: 85% of these buyers think more highly of a vendor when an AI includes it in its answer. Being cited without a click is therefore not nothing. The name is read, and it weighs in the choice, even if no measurement tool records it.
What to do
First, measure what remains: visits coming from AIs can be counted separately in Google Analytics, and another page in this section explains how. Next, look at your customers' questions rather than your keywords: on which questions should your company be named, and who is named in its place?
Finally, give AIs reasons to cite you. They build their answers from the pages they find, most often somewhere other than your site: Muck Rack, across more than 25 million links cited by ChatGPT, Claude and Gemini, counts 84% of sources that brands do not control (May 2026 report). The more serious sources there are that describe what you do, the more chance your name has of entering the answer, even when nobody clicks.
Where FaireDuBruit stands
FaireDuBruit, the publisher of this page, is a distribution service. Before each distribution, you receive an email, then an article is written and published on your site and on three specialist media, in your language and in English, one to three times a week depending on the plan. It costs from €299 excl. VAT per month, with no commitment, plus a setup of €690 excl. VAT paid once.
Its limits should be known. It measures neither your clicks nor your sales, and does not track your citations: a monthly report only says which AIs read your articles on our media. Those on our media are labeled as sponsored. No citation by an AI is promised.
What we do not know
Pew's figures cover U.S. adults, in March 2025, and Google's AI summaries only; these answers may have changed since. Adobe's cover retail in the United States. G2's cover B2B software buyers. None of these measurements counts lost customers.
Sources
- Pew Research Center, Athena Chapekis and Anna Lieb — “Google users are less likely to click on links when an AI summary appears in the results”, .
- Adobe, Vivek Pandya — “AI traffic now generates 83% higher revenue per visit for U.S. retailers…”, .
- G2, Tim Sanders — “The Answer Economy”, .
- Muck Rack — “What is AI Reading?” (Generative Pulse), .
- FaireDuBruit measurements, from September 4 to 16, 2026 — see “How we measured”.