6 November 2026

Quick answer ChatGPT and similar assistants recommend local businesses using a mix of map and directory data, web search results, reviews and mentions across trusted sites. To be included, keep your Google and Bing profiles complete and consistent, earn detailed reviews, publish clear service and location pages, and be mentioned on reputable local and industry sites. Last updated: November 2026 More people now ask an AI assistant questions like “Can you recommend a reliable plumber in Bristol?” or “Which dentist near me does Invisalign?”. The reply often names two or three businesses, and customers frequently contact those first. Nobody outside the AI companies knows the exact recipe, but the behaviour of these tools is consistent enough to draw practical conclusions. They rely heavily on data that already exists online: local listings, reviews, websites and mentions. This guide explains how recommendations are formed, which signals seem to matter most, a step-by-step plan for UK businesses, and how to test whether you are being recommended. In this guide Where AI assistants get local business information The signals that appear to matter most What assistants cannot do Why detailed reviews outperform generic ones A step-by-step plan for UK businesses What a recommendation actually looks like Create content that assistants can quote How to test whether you are being recommended Common mistakes Do you need to do anything special for AI? FAQs Where AI assistants get local business information When browsing or search is enabled, assistants query search engines and mapping data and then summarise the best matches. Many use Bing or Google results, while integrations with maps, review platforms and directories provide business details such as address, opening hours and ratings. Without live search, an assistant relies on what it learned during training, which tends to favour well-known brands and widely mentioned businesses. Smaller businesses are more likely to appear when live search is on and their online information is clear and current. The signals that appear to matter most Complete, consistent profiles on Google, Bing and Apple, with correct categories and services. Volume, recency and detail of reviews, particularly reviews that mention specific services and areas. Clear website pages that state what you do, where you do it and for whom. Mentions and links from local press, directories, associations and partner sites. Positive sentiment across the web, including forums and social platforms. No single signal dominates. Businesses that are strong across several are the ones that tend to be named. What assistants cannot do They do not have a secret list of paying advertisers, and you cannot buy a recommendation directly. They also make mistakes, such as quoting old phone numbers or merging two similar businesses. That is why consistent details and a strong online footprint matter, since they reduce the room for error. Treat AI visibility as an outcome of overall reputation and information quality, not a standalone project. Why detailed reviews outperform generic ones A review that says “Great service, five stars” gives an AI little to work with. A review that says “Dave fitted our new combi boiler in Headingley in one day, tidy and fairly priced” provides services, place and sentiment in one sentence. Encourage customers to describe what you did, gently, without scripting their words. A step-by-step plan for UK businesses Start with the foundations, because they feed every assistant. Claim and complete your Google Business Profile, Bing Places listing and Apple Business Connect entry. Use your exact trading name, choose accurate categories, list services and keep hours up to date. Make sure your website, profiles and directories use identical name, address and phone details. Write a clear page for each core service and each area you serve, opening with a plain summary. Add LocalBusiness and FAQ schema to key pages. Build a steady flow of detailed reviews and reply to them. Earn mentions from local press, associations, suppliers and community sites. What a recommendation actually looks like Suppose a customer asks, “Who is a good family dentist in Cardiff who offers payment plans?”. A typical assistant response lists two or three practices, describes what each offers, mentions ratings or review themes, and often links to the sources it used. The practices named usually share traits: a well-completed profile, plenty of detailed reviews mentioning family care, and a website page that explains treatments and finance options in plain language. A practice with a vague website, a handful of reviews that say little and a different phone number on each directory is much less likely to be named, even if its clinical work is excellent. The assistant cannot verify what it cannot find, so it recommends businesses where the evidence is clear. This is encouraging for small firms, because evidence is something you can build. A page that explains payment plans, reviews that mention the experience of families, and a consistent presence across profiles are all within your control. Treat each one as a piece of evidence that an assistant, and a human customer, can quickly check. It also explains why updating old information matters. Assistants may quote pricing, opening hours or services from outdated pages. Reviewing your key pages and listings every quarter keeps that evidence current and prevents you losing recommendations because of stale details. Create content that assistants can quote Assistants favour pages that answer a question cleanly. For each service, write a short summary that states what you offer, who it is for, typical costs or ranges, and the areas covered. Follow with sections on process, timescales and common questions. Pricing transparency helps. Many people ask assistants about cost, and a page that gives honest ranges is more likely to be used than one that says only “contact us for a quote”. If you cannot publish exact prices, explain what affects them and give a typical range. How to test whether you are being recommended Make a list of 15 to 20 realistic prompts customers might use, such as “best electrician in Sheffield for a rewire” or “recommended accountant for contractors in Birmingham”.