The New Shortlist: How AI Platforms Are Quietly Deciding Which Estate Agencies Get Seen
Every housing market runs on visibility, and for decades that visibility was shaped by listings pages, referrals, and search rankings. A new dataset suggests the gatekeeper has quietly changed. A study from research firm Schutle, built from 10,560 AI responses across ChatGPT, Gemini, Claude, and Google AI Overviews, found that when buyers, sellers, landlords, and investors ask an AI platform for an estate agency recommendation, they are typically handed a shortlist of just three to five names. Everyone else, regardless of reputation, may simply not exist in that conversation.
This is the kind of signal I find most interesting, not because it is dramatic, but because it is structural. The research mapped real markets: Knight Frank held 63.8% visibility in Central London prompts, Hurford Salvi Carr led Canary Wharf and Docklands queries at 30.3%, and in Marbella, Engel & Völkers appeared in 96.6% of responses aimed at German buyers, while independent firm Bromley Estates Marbella outperformed larger international names among British buyers at 81.2%. These are not small variances. They describe an entirely new commercial layer sitting between a business and its customer, one built from probability and training data rather than paid placement or word of mouth.
What should catch the attention of anyone tracking property intelligence is the platform inconsistency. Dubai-based Harbor Real Estate showed up in 95 ChatGPT responses and 15 on Gemini, yet did not surface at all in Google AI Overviews. Visibility, in other words, is not a single score. It is fragmented across systems that weigh evidence differently, and a business can be simultaneously prominent and invisible depending on which assistant a customer happens to open.
The mechanics behind these outcomes matter more than the rankings themselves. Schutle’s founder, Jaimie Beers, noted that AI systems draw on operating history, independent reviews, professional standards, transparency, local expertise, language capabilities, and specialist market experience when forming a recommendation. None of that is new criteria for a good agency. What is new is the requirement that this evidence be structured and machine readable, not just true.
An agency may offer an excellent service, but if its public digital footprint does not clearly demonstrate that capability, the AI may have little evidence on which to recommend it.
That distinction between having a capability and proving it in a format a model can parse is the real story here. Property intelligence has always been about turning scattered signals into a clear picture, and this data suggests the audience for that clarity now includes the algorithms sitting between a business and its next client. Agencies that treat their digital footprint as a dataset to be organized, rather than a brochure to be published, are the ones showing up when it counts. For an industry still adjusting to AI as an intermediary rather than a tool, that is a pattern worth watching closely.
Source: BusinessMole, “AI’s top estate agent picks and the reasons behind them uncovered in latest study”
