--- title: How AI changed the way customers choose you date: 2026-07-02T00:00:00+08:00 author: Jessica Kaitse canonical_url: "https://remedystudio.com.au/insights/why-ai-isnt-recommending-your-business" section: Insights Articles --- July 02, 2026 ## How AI changed the way customers choose you AI sits between your customers and your website. People research a problem, get an answer, and weigh up who to trust before they ever land on your site or make contact. The traffic that has fallen is mostly the early, informational kind. The people who become customers are still there, but they continue to do their homework in private first as they've always done. AI just gives them a better convenient tool than your website to be able to do this. We saw this with a client recently who runs a niche business, the kind where only a handful of operators compete for the same work. For years their site quietly did its job, then over a few months the traffic slid and kept sliding. Their read on it was blunt: *"We worked so hard to get up there on Google, and it just vanishes overnight with AI."* They are not wrong, and they have not done anything wrong, the goal posts have shifted. ## Where did the website traffic go? It went to the answer, before anyone reaches a website. Two shifts are happening at once. 1. Google now shows an AI summary above the results, and clicks drop when it appears. [Pew Research](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) found people clicked through to a website 8% of the time when an AI summary was present, against 15% when it was not, close to half, and only 1% clicked a link inside the summary itself. 2. The second shift gets less attention. Many people have stopped using a search engine for these questions at all and ask an assistant directly. [ChatGPT](https://www.demandsage.com/chatgpt-statistics/) now has around a billion users and handles more than two billion queries a day, used most for the open-ended questions people once typed into Google. Some fields feel this more than others. [SE Ranking](https://www.americanbar.org/groups/law_practice/resources/law-practice-magazine/2026/march-april-2026/ai-powered-search-visibility-for-law-firms/) found legal searches return an AI answer around 78% of the time, one of the highest rates of any category, so a person with a question about their options gets a workable answer and never lands on a firm's site. ## Does falling traffic mean people stopped needing you? No. People are changing when they make contact, not whether they need you, so a drop in traffic is not the same as a drop in demand. In [Rev's 2026 AI Legal Advice Index](https://www.rev.com/blog/ai-legal-advice-index), about two-thirds of people said they had used an AI chatbot for legal help, but mostly to clarify a term or research their situation before speaking to someone. **For anything with real consequences, they still want a person.** AI absorbs the early, informational part of the journey, the part that used to bring people to the website first. The serious enquiries still exist, but they just do not show up as traffic the way they used to. ## Why do people research with AI before they get in touch? Because it is private, fast and (mostly) free. A [national survey of 1,000 US adults](https://goodmenproject.com/featured-content/the-ai-trust-gap-42-would-ask-ai-before-calling-a-lawyer/) found 42% would turn to AI before contacting a lawyer, and the reason people gave was that AI feels fast, private and free. Asking it costs nothing, reveals nothing, and never makes someone feel sold to, which matters most for anything sensitive. They are not going it alone. They want to be informed, and to quietly check you out, before they involve anyone. [Martindale-Avvo](https://www.martindale-avvo.com/blog/how-do-clients-research-and-find-their-attorneys/) found more than 90% of legal consumers research their issue before contacting a lawyer, most visit at least two websites, and close to half use review sites and directories along the way. They are weighing you up well before you know they exist. ## Which searches can your own website win? The informational ones. For questions about what a process involves, what someone's options are, or how something works, your own pages can be the source an AI uses, as long as they are structured well and genuinely useful. What it draws from is usually a cluster of focused, linked pages around a topic rather than one page trying to cover everything. [Yext's analysis](https://www.digitalapplied.com/blog/ai-search-citation-ranking-factors-2026-data-study) of 6.8 million AI citations found sites with topic clusters were cited around three times more often than single-page competitors. The "who should I use" search is the one your own site rarely wins. When an AI recommends a business, it leans on sources it treats as independent. [Muck Rack's May 2026 study](https://muckrack.com/blog/what-is-ai-reading-may-2026) of more than 25 million AI citations found about 84% came from earned media and third-party sources rather than brands' own websites. Martindale-Avvo's data points the same way, with review sites and directories playing a large part in how people build a shortlist. Part of being chosen sits off your website entirely, through being reviewed, listed and mentioned. ## How do you stay visible when AI answers first? It depends on where you are starting, and it is not always a rebuild. Two jobs, matching the two kinds of search. To be the source an AI uses for the informational questions, build depth where it counts: focused, well-structured pages, grouped into clusters around the topics people ask about, linked together and kept current, so an assistant can lift a clean answer and attribute it to you. Thin or scattered pages give it nothing to quote. To be named in the "who should I use" answers, work the half that sits off your website. Keep your Google Business Profile accurate and active, get your listings consistent across the directories that matter in your sector, and build a steady flow of genuine reviews. Those are the independent signals an AI leans on when it decides who to recommend. A rebuild only comes into it when the site is too old to do either of those well. If it is the better part of a decade old and awkward to change, it is probably slower and less secure than it should be, and built for a version of search that no longer exists. In that case the rebuild is what makes the rest possible. If the site is reasonably new and easy to update, you do not need one, you need the two jobs above. A note on the numbers. The studies here are US-based, applied to our context as directional, since Australian data on this is still thin. And there are honest limits: you cannot see what people type into ChatGPT the way you could read Google search terms, and even Google's own AI clicks are hard to separate in analytics, so some of this is informed reading rather than exact data. The judgement still comes from people who know the work. None of this brings back the old version of search, and it does not have to be pure loss. The informational traffic has gone. The person quietly deciding whether you are the right choice has not. Being the business they pick is now half a website job and half a reputation one. --- **Sources:** [Pew Research Center (2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) · [SE Ranking, via the ABA (2026)](https://www.americanbar.org/groups/law_practice/resources/law-practice-magazine/2026/march-april-2026/ai-powered-search-visibility-for-law-firms/) · [Rev AI Legal Advice Index (2026)](https://www.rev.com/blog/ai-legal-advice-index) · [The AI Lawyer Trust Gap survey (2026)](https://goodmenproject.com/featured-content/the-ai-trust-gap-42-would-ask-ai-before-calling-a-lawyer/) · [Martindale-Avvo Legal Consumer Report](https://www.martindale-avvo.com/blog/how-do-clients-research-and-find-their-attorneys/) · [Yext AI Citation Study, via Digital Applied (2026)](https://www.digitalapplied.com/blog/ai-search-citation-ranking-factors-2026-data-study) · [Muck Rack, What is AI Reading? (May 2026)](https://muckrack.com/blog/what-is-ai-reading-may-2026) · [ChatGPT usage statistics (2026)](https://www.demandsage.com/chatgpt-statistics/) --- ## About the author ![REMEDYNOV2540 }}](https://remedystudio.com.au/assets/uploads/images/RemedyPhotos2025/REMEDYNOV2540.jpg) ![REMEDYNOV2540](data:image/gif;base64,R0lGODlhwgEBAPAAAP///yZFySH5BAEAAAEALAAAAADCAQEAAAIThI+py+0Po5y02ouz3rz7D4ZTAQA7) **Jessica Kaitse** is a co-founder of Remedy Studio, a Perth web design and development studio that builds and looks after websites for schools, not-for-profits and healthcare organisations. She has worked on websites for more than 25 years and in SEO for 20. She focuses on strategy, content, and building sites that work now and hold up later, the thinking behind Remedy's vision: **build the present, own the future.** ### Share this article Twitter/X LinkedIn Facebook Copy link ### Written by Remedy We're a Perth-based web design and development agency helping schools, nonprofits, and businesses build digital infrastructure that lasts. We believe great design should be accessible to organisations doing important work. [Can AI build your website? ](https://remedystudio.com.au/insights/can-ai-build-your-website-what-decides-the-result)