Your Customers Already Ask AI First

For most of the past twenty years, a customer who wanted a plumber, a dentist or an accountant typed a few words into a search engine and chose from a list of links. The list had an order, and a business could work to move up it. That arrangement is still there, but it is no longer the first thing many people see.

In February and March 2026, Pew Research Center asked 5,119 American adults about their use of AI. About half now use AI chatbots, up from a third in 2024. The most common reason, given by 42%, was to search for information. A separate finding concerns Google itself: six in ten adults say they read the AI summaries that now appear at the top of search results.

The consequence is simple to state. A growing share of customers form a view of your business from an answer written by a machine, before they have visited your site or picked up the phone. The rest of this piece looks at how that changes discovery, how it changes expectations once contact is made, and what a small business can do about both.

The clearest evidence for local businesses comes from BrightLocal’s Local Consumer Review Survey 2026, a SurveyMonkey panel of 1,002 American adults. It asked where people look for recommendations when choosing a local business. Use of ChatGPT and other generative AI tools rose from 6% a year earlier to 45%, which made AI the third most popular source of recommendations. Over the same period, the share using Google fell from 83% in 2025 to 71%.

Google remains the larger channel by a wide margin, so this adds a route without closing the old one. But a jump from 6% to 45% in twelve months is a large movement, and it describes behaviour in the least glamorous category of commerce: the shop, the clinic and the tradesman down the road.

Trust follows use. In the same survey, 40% of consumers said they trust AI platforms to provide business recommendations, against 32% who do not, and 42% said they trust AI platforms as much as traditional reviews. The panel is American, and the equivalent figures for Bulgaria or Spain may differ. The direction is unlikely to stop at the Atlantic.

Part of the shift happens inside Google. Pew Research Center tracked the browsing of 900 American adults during March 2025 and compared visits where Google showed an AI summary with visits where it did not. When a summary appeared, users clicked a traditional result link in 8% of visits. Without one, they clicked in 15%. Links inside the summary itself were clicked in 1% of visits. Users ended their browsing session entirely on the page in 26% of cases when a summary was present, against 16% when it was absent.

Pew also found that the summaries were not yet universal: 18% of the 68,879 searches in the study produced one. These are 2025 figures, and Google’s rollout has continued since.

Bain & Company reached a similar picture from a different direction. Its February 2025 analysis, drawing on a December 2024 survey of 1,117 consumers, found that about 60% of searches on traditional search engines now end without the user going on to another site, and that 80% of consumers rely on such zero-click results at least 40% of the time. Bain estimated that this was reducing organic web traffic by 15% to 25%.

When the answer is written for the customer, the website becomes source material as much as a destination. What the AI says about a business depends on what it can find: the opening hours on the website, the services listed, the reviews, the entries on maps and directories. If those disagree with each other, the summary may be wrong, or the business may be left out.

This is the part owners can influence. A site that states plainly what the business does, where, for whom and at what price gives an AI system accurate material to quote. So does a consistent name, address and phone number across directories, and a steady supply of recent, specific reviews. BrightLocal found that 82% of consumers now read AI-generated review summaries, and that 23% would rely on them alone to decide. A review that was never written cannot be summarised.

None of this is exotic. It is the older discipline of keeping a business’s public record accurate, now with a reader that never gets tired and does not click through to check.

A person who reaches out after an AI conversation tends to arrive further along. They have asked a question, seen a short list and often compared two or three names. The enquiry that follows is closer to a decision than a browse, which makes the first reply matter more than it used to.

It also means fewer people reach the website at all. With clicks falling where summaries appear, the first real contact may be a message on WhatsApp, a call or a form, sent from a phone, often outside office hours. The front door is not only in a different place. It is open at different times.

Zendesk’s CX Trends 2026 report states that 88% of customers expect faster response times than they did a year ago and that 74% of consumers now expect customer service to be available 24/7. These are expectations of service in general, and they do not measure how quickly a particular business loses a sale.

Measured behaviour suggests that most businesses are some distance from those expectations. Hennessey Digital, a marketing agency, studied how more than 1,300 American law firm websites handled online enquiries and published the results in June 2025. A quarter of the firms replied in under five minutes, up from 13% in 2021. Another 26% still failed to respond at all. Law is one profession and the sample is American, but the pattern of a fast minority and a silent minority is familiar from many trades.

A frequently cited 2011 study in the Harvard Business Review, by Oldroyd, McElheran and Elkington, reached the same conclusion fifteen years ago, stating that most companies are not responding nearly fast enough. Zendesk’s figures suggest that expectations have only risen since.

Put the two shifts together. A customer asks an AI tool in the evening, receives a short list, and sends a message to the first name that looks credible. If the business answers at nine the next morning, the customer has had the whole night to message the second name. The reply is not late by the owner’s standards, but it is slow by the customer’s.

The practical response has two parts. First, check what the AI tools currently say about the business: ask them the questions a customer would ask, and correct the sources behind any errors, starting with the website and the map listing. Second, decide who or what answers an enquiry in the first few minutes, including at night and at weekends. For many small firms that will be a simple automated first response that confirms the enquiry, answers the common questions and books a time, with a person taking over from there. The aim is modest: that no enquiry waits unanswered, whenever and wherever it arrives.