AI shortlists only three to five estate agents, study finds

What the study found

Schutle, an AI search visibility company, published research on 18 September 2026 on how ChatGPT, Gemini, Claude and Google AI Overviews recommend estate agencies to buyers, sellers, landlords and investors. The study analysed 10,560 AI responses across 116 prompts covering London, Dubai and Marbella, run repeatedly between 1 June and 31 July 2026. The central finding is one of narrowness. AI platforms typically recommended only three to five agencies in each response, meaning that a question which might once have produced a page of search results now produces a shortlist small enough to read in a few seconds. For an agent, being absent from that shortlist is not a ranking problem in the familiar sense. It is an absence from the conversation altogether.

The companies named changed considerably with the customer's location, nationality, requirements and choice of AI platform. A buyer asking about a two bedroom flat in E14 did not receive the same names as a landlord asking about yield in prime central London, and the platform used mattered as much as the question asked. Schutle's method counted only firms clearly recommended or presented as suitable shortlist options. Incidental mentions, citations, comparisons and negative references were excluded, so the visibility figures reflect positive inclusion in an answer rather than any appearance of a brand name in the text. That distinction matters because an agency can be discussed by an AI assistant without being recommended by it, and the study treats those two outcomes as different.

Who wins in London

Knight Frank led Central London recommendations, appearing in 63.8% of relevant responses. That figure places the firm in roughly two of every three answers where a central London agency was recommended at all, a level of presence that leaves comparatively little room for others in a shortlist of three to five names. The result is not a statement about market share, transaction volumes or client satisfaction. It is a measure of how often a single firm is surfaced by AI platforms when a user asks for an agent in central London. Even so, the concentration is notable: in a format that names only a handful of firms, a 63.8% appearance rate implies that many answers contain Knight Frank alongside a small number of others, and some answers contain it alone.

The picture changes sharply by location within the capital. Hurford Salvi Carr led Canary Wharf, Docklands and E14 prompts with 30.3% visibility. That is a lower figure than Knight Frank's central London result, which suggests the Docklands recommendation set is more fragmented, with no single firm dominating the way Knight Frank does in the centre. The shift between the two areas illustrates the study's broader point: recommendations changed considerably with the customer's location, nationality, requirements and choice of AI platform. An agency with strong visibility in one postcode may have little or none in an adjacent one, and a buyer's stated nationality or budget profile can alter which names appear. For firms operating across London, the practical implication is that AI visibility is not a single national score but a patchwork of local outcomes.

How it was measured and what it does not show

The study covered 116 prompts across London, Dubai and Marbella, run repeatedly between 1 June and 31 July 2026, producing 10,560 AI responses in total. Four platforms were included: ChatGPT, Gemini, Claude and Google AI Overviews. Prompts were designed to reflect the questions buyers, sellers, landlords and investors actually ask, and the responses were then assessed for which agencies were clearly recommended or presented as suitable shortlist options. Incidental mentions, citations, comparisons and negative references were excluded from the counts. The repetition of prompts over two months allowed the researchers to observe how stable or variable the answers were, though the published findings focus on the visibility percentages rather than on run to run consistency.

Schutle said the results measure AI recommendation visibility and do not assess service quality, transaction volumes, achieved prices or customer satisfaction. The company also said that AI answers change as models and source material update, which means the figures describe a period rather than a permanent state of affairs. Schutle is an AI search visibility company, so the research is published by a vendor with a commercial interest in the field it measures. That does not invalidate the findings, but it is a relevant fact for anyone reading them. The study does not claim that appearing in an AI answer causes a customer to instruct an agent, nor that absence from an answer prevents a customer from finding a firm by other means. It measures what the platforms said during a defined window, using a defined method, and reports the resulting percentages.

Why it matters beyond property

Schutle's conclusion is that AI is creating a new commercial consideration set, and that companies not included in an AI-generated answer may never reach the customer's original shortlist. The phrase consideration set is borrowed from marketing research, where it describes the small group of brands a buyer actively considers before choosing. What the study describes is the formation of that set inside an AI answer, before the customer has visited a search engine results page, a comparison site or an agency's own website. If the shortlist is three to five names, then the commercial contest is increasingly about entry to that list rather than prominence within a longer set of options. A firm that would have ranked eighth on a search page might once have been found by a determined buyer scrolling further. In an AI answer, there is often no further to scroll.

Kael Tripton context, stated as analysis, notes that the FCA's April 2026 consumer survey found 16% of UK consumers already use AI for money tasks, so the shortlist effect the study describes is reaching consumer finance as well as property. The same structural feature applies: an AI assistant asked to suggest a mortgage broker, a savings account or an adviser will tend to name a small number of options, and the firms outside that set are not part of the conversation the customer is having. For estate agents, the study offers a specific measurement of a general shift. For anyone watching how consumers find professional services, the property data provides an early and unusually detailed example of how AI answers are shaping which companies get considered at all.

Source: Schutle research release.

Related coverage on Kael Tripton: Estate Agent Insurance UK 2026: PI, Cyber and Professional Cover, Bookkeeping Software For Real Estate Agents: UK Business Software Guide, Do You Pay Stamp Duty When Selling A House, Letting Agent Software UK 2026: Lettings CRM and Compliance Guide, Real Estate Accounting Software UK: Letting Agents and Landlords.

Frequently asked questions

How many AI responses did the study analyse?

Schutle analysed 10,560 AI responses across 116 prompts covering London, Dubai and Marbella, run repeatedly between 1 June and 31 July 2026.

Which AI platforms were included?

ChatGPT, Gemini, Claude and Google AI Overviews.

How many estate agencies did AI platforms typically recommend?

AI platforms typically recommended only three to five agencies in each response.

Which agency led Central London recommendations?

Knight Frank led Central London recommendations, appearing in 63.8% of relevant responses.

Which agency led Canary Wharf, Docklands and E14 prompts?

Hurford Salvi Carr led Canary Wharf, Docklands and E14 prompts with 30.3% visibility.