# AI search for estate agents: how to get recommended by ChatGPT > How buyers and sellers use ChatGPT, Google AI and Perplexity to pick agents, what those engines read, a checklist for agencies and a simple way to measure it. URL: https://www.formule.io/uk/blog/ai-search-for-estate-agents Language: en Publisher: Formule (Formule Labs Limited) Updated: 2026-11-03 Sources checked: 2026-11-03 **Buyers, sellers and landlords now put the same question to ChatGPT, Google AI Overviews, Perplexity and Copilot that they once typed into a search box: who is a good estate agent near me, and how do these two agencies compare. An agency gets recommended when the engines can find clear, consistent, well reviewed information about it in the places they read. No engine publishes a switch that buys a recommendation, so the work is ordinary and checkable.** This guide explains how people use AI to find an agent, what the engines draw on according to their own documentation, a checklist an agency can finish in a few weeks, and a simple method to measure whether anything changed. It is honest about what is unproven. For the wider picture of AI in an agency, start with [AI for estate agents: what works today, what doesn't, a 1-week plan](https://www.formule.io/uk/blog/ai-for-estate-agents). For the UK rules around AI in an agency, read [AI for estate agents in the UK: leads, DMCC, AML and data rules](https://www.formule.io/uk/blog/ai-for-estate-agents-uk). ## Buyers and sellers now ask AI to shortlist agents The questions come in three shapes. A seller asks which agents sell family homes in their town and what they charge. A landlord asks for a lettings agency that handles tenant referencing and deposits. A buyer asks which agency to register with for a two bedroom flat near a station. The answers read like a shortlist with a sentence of reasoning for each name, often with links. That changes what an agency is competing for. In a classic results page, ten links share the click. In an AI answer, a handful of names are mentioned and the rest are invisible. Being left out of the answer costs you the enquiry before the person has visited a single website. It also changes who reads your pages. The reader is increasingly a machine summarising them for someone else, so pages that state facts plainly, such as which areas you cover, what services you offer and who works there, travel better than pages that rely on photographs and slogans. ## Each engine builds its answer from a different mix of sources The four engines differ, and their own documentation says how. ### Google AI Overviews and AI Mode Google's page on [AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) says these features use a technique called query fan-out, running several related searches at once, so the links shown can be wider and more varied than in classic search. It also says a page must be indexed and eligible to appear in Search with a snippet, and that there are no additional technical requirements. Google states you do not need new machine-readable files, AI text files or special schema markup for these features. In plain terms, the road into Google's AI answers is the road into Google Search. For local searches, Google describes how its local ranking works. Its [Business Profile help](https://support.google.com/business/answer/7091) says local results are based mainly on relevance, distance and prominence, that more reviews and positive ratings can help local ranking, and that businesses with complete and accurate information are more likely to show up. It also says there is no way to pay for a better local ranking. The page does not say how this feeds Google's AI answers, so treat it as the best available guide to the local signals. ### ChatGPT OpenAI documents its crawlers on its [bots page](https://developers.openai.com/api/docs/bots). OAI-SearchBot is the one used for ChatGPT search results. GPTBot is a separate crawler for content that may be used to train models. ChatGPT-User fetches pages when a person asks a question and is not an automatic crawler. OAI-SearchBot and GPTBot are controlled separately in robots.txt, so an agency can allow search visibility and still disallow training. OpenAI says that robots.txt rules may not apply to ChatGPT-User because a person initiates those visits, and says a site that opts out of OAI-SearchBot will not appear in ChatGPT search answers, though it can still show as a navigational link. ### Perplexity Perplexity's [crawler guide](https://docs.perplexity.ai/guides/bots) says PerplexityBot surfaces and links sites in Perplexity search results and is not used for training foundation models. Perplexity-User visits a page when a person asks a question and generally ignores robots.txt because the person requested the fetch. ### Copilot We have not found Microsoft documentation that sets out how an agency can influence Copilot answers, so this guide makes no claim about how Copilot picks agencies. Treat it as one more reader of the open web and include it in your checks. ## The sources behind a recommendation are mostly not your own site Google and the others do not publish a list of the sites they favour, so what follows is an inference from how the documentation works. Treat it as a map of where to look, not a promise. - **Your own pages.** Area pages, service pages, team pages and valuation pages are the only sources you fully control. An engine can only repeat what it can read. - **Google Business Profile and reviews.** For anything phrased as near me or in a town, the profile and its reviews carry the local signals Google documents. - **Portals.** Rightmove, Zoopla and OnTheMarket carry live listings, so an answer describing the stock an agency handles can draw on them. - **Local press and awards.** A named agency in a local newspaper, an awards shortlist or a trade article gives an engine an independent mention to cite. - **Directories and comparison pages.** Trade bodies, redress scheme member lists and agent finders confirm that you exist and are regulated. The pattern is that an engine trusts a fact more when two or three independent places state it the same way. That is why consistency matters more than volume. ## A practical checklist for an agency Work through these in order. The first four are the foundation. The rest add detail. - **Make the site crawlable and indexed.** Confirm in Google Search Console that your main pages are indexed. Put important facts in text, not only in images or PDFs, because Google's documentation lists making important content available in textual form as a best practice for its AI features. - **Write area and service pages.** One page per town or neighbourhood you cover, and one per service: selling, letting, valuation, property management. Each should say who it is for, what you charge or how fees work in plain words, how long the process takes and who to contact. Avoid copying the same paragraph across areas. - **Keep name, address and phone identical everywhere.** Use one spelling of the agency name, one address format and one number on the website, Google Business Profile, portals, social profiles and directories. Fix the old trading name that still appears on one directory. - **Claim and complete the Google Business Profile.** Choose accurate categories, add opening hours, service areas, a description and photographs, and keep them current. Google says businesses with complete and accurate information are more likely to show up in local results. - **Earn reviews honestly.** Google's [review policy](https://support.google.com/business/answer/2622994) lets you ask customers for genuine reviews but forbids offering anything in return, discouraging negative reviews or asking only happy customers. Ask every client after completion, reply to each review in your own words, and never pay for or write one yourself. - **Publish clear about and team pages.** Name your agents, their roles, your redress scheme and your years trading. Engines and buyers both look for who stands behind the name. - **Add structured data only where Google documents it.** Google's [local business structured data guide](https://developers.google.com/search/docs/appearance/structured-data/local-business) lists the properties it reads, such as name, address, telephone, opening hours and URL, and states it does not guarantee that any feature will show. Keep the markup consistent with what the page visibly says. Google says review and aggregate rating markup is only recommended for sites that capture reviews about other local businesses, so do not add star ratings for your own agency by default. - **Allow the AI search crawlers.** Check your robots.txt. If it blocks OAI-SearchBot or PerplexityBot, ChatGPT search and Perplexity are told to stay away. Allowing a search crawler is a separate decision from allowing a training crawler. Google also notes that robots.txt is [not a mechanism for keeping a page out of Google](https://developers.google.com/search/docs/crawling-indexing/robots/intro), so use the right control for what you actually want. Ask your web provider to show you the live file, because plugins and security settings often add blocks nobody remembers. - **Earn independent mentions.** Offer a market comment to the local paper, enter the regional awards, and make sure the portals and your redress scheme list you correctly. - **Treat llms.txt as optional.** An llms.txt file is a proposed plain text summary of a site for AI readers. Google says outright that its AI features need no special AI text file, and we have seen no engine document that it reads one to choose agents. Adding one costs an hour and is harmless. Do not expect it to move your results. ## A fixed prompt panel shows whether you are named You cannot see your ranking inside an AI answer the way you see a position in a results page, and a single check proves nothing because answers vary from one run to the next. A fixed panel gives you a series you can compare over time. The method needs no special tools. - **Write the panel once.** Choose around twenty questions a real client would ask, phrased as people phrase them: agents in your town, comparisons with your two nearest rivals, lettings versus sales, fees and valuations. Include a few that name your agency directly. - **Freeze the wording.** Never edit a question. Add new ones at the end with new numbers so older results stay comparable. - **Use the same engines each time.** ChatGPT with search on, Google with its AI Overview or AI Mode, Perplexity and Copilot, each in a clean session with no personalisation. Record where the engine says it is located. - **Score every answer on the same columns.** Date, engine, question number, whether your agency is named, whether your website is cited as a source, your position among the names given, the competitors named, and any wrong statement about you. - **Run it on a schedule.** Once a month in the same week, by the same person. - **Read the wrong statements first.** An engine that states your fees or your areas incorrectly is usually reading an old page. That is a fix you can make. Two cautions. Answers are not stable, and a change of one name in or out is noise. Only a pattern over several months means something. And a panel measures visibility, not enquiries, so keep reading your enquiry sources alongside it. Build in both generic questions and questions that name a specific brand or comparison, because the answers can differ. ## What is unproven Be sceptical of anyone selling certainty here. Nobody outside the engines knows the exact weighting of reviews, press, portals and your own pages, and the weighting changes. There is no published evidence that llms.txt influences recommendations. Schema markup helps Google understand a page, but Google states it guarantees no feature. Reviews matter for local ranking according to Google, yet how far they matter inside a generated answer is not documented. What does hold up is that each step on the checklist makes the agency easier for a person to trust, which is the part that survives any change in the engines. ## Answering the enquiry that AI sends you Better visibility only pays if the enquiry is answered quickly. People who arrive from an AI answer often come by WhatsApp or a short form, with a specific question in hand. Really, Formule's AI agent, answers enquiries on WhatsApp and drafts replies that the agent approves before anything goes out, so a late evening message gets a considered answer waiting for you in the morning. The step by step process is in [AI lead follow-up on WhatsApp: a fast, personal reply to every enquiry](https://www.formule.io/uk/blog/ai-lead-follow-up-whatsapp), and a comparison of other tools by job is in [Best AI tools and software for estate agents in 2026, by job](https://www.formule.io/uk/blog/best-ai-tools-for-estate-agents). If you want to see Really on your own enquiries, [start free](https://app.formule.io/signup?lang=en&utm_source=blog&utm_medium=organic&utm_campaign=ai-search-agents&utm_content=en) and it sets itself up in one chat. ## A four week plan - **Week one.** Check indexing, check robots.txt, claim and complete the Google Business Profile, fix name and address mismatches. - **Week two.** Write or rewrite the area and service pages for your two strongest towns, and publish the team page. - **Week three.** Start the review routine for every completed instruction, and request listings corrections on portals and directories. - **Week four.** Build the prompt panel, run it once as your baseline, and write down every wrong statement to fix. Repeat the panel the following month. The first run is a baseline, not a verdict. ## Sources - https://developers.google.com/search/docs/appearance/ai-features - https://developers.openai.com/api/docs/bots - https://docs.perplexity.ai/guides/bots - https://support.google.com/business/answer/7091 - https://support.google.com/business/answer/2622994 - https://developers.google.com/search/docs/appearance/structured-data/local-business - https://developers.google.com/search/docs/crawling-indexing/robots/intro