AI visibility research
Which Central and East London estate agents are ChatGPT, Gemini and Google AI recommending, and why?
2,595
AI-generated responses analysed
51
Commercially specific prompts
3
AI platforms
ChatGPT, Gemini, Google AI Overview
Discover what 2,595 AI-generated responses reveal about which estate agencies are being recommended across Central and East London, how the results change between sales and lettings, and why different agencies lead in different markets and locations.
Our research examines 51 commercially specific prompts across three AI platforms, exploring the agencies recommended to sellers and landlords, the differences between local markets, and the reasons AI gives for its recommendations.
From Knight Frank and Savills to Dexters, Foxtons and independent local specialists, the results reveal a competitive landscape where established reputation, local expertise and service-specific strengths all feature in AI-generated recommendations.
Knight Frank led Central London with a 63.8% AI recommendation share, closely followed by Savills at 60.9% and Dexters at 60.2%. The results show strong competition between established agencies, with all three appearing prominently across the questions analysed.
Foxtons led general East London with 37.0%, narrowly ahead of Keatons at 35.2% and Savills at 33.7%. The results demonstrate how local specialists can compete alongside major national brands, and how recommendation leaders change between different parts of London.
Savills led the sales and seller outcomes analysis at 49.4%, while Dexters ranked first for both lettings and property management at 34.2% and reputation and customer service at 41.1%. The findings show that AI recommends different agencies depending on what the customer is looking for, rather than consistently favouring one overall market leader.
Knight Frank leads Central London (63.8%), Foxtons leads general East London (37.0%) and Hurford Salvi Carr leads Docklands (30.3%).
Savills leads seller outcomes (49.4%), while Dexters leads lettings and management (34.2%).
Reputation appears in 77.2% of seller responses, marketing reach in 75.0% and valuation expertise in 61.4%.
Hurford Salvi Carr leads Docklands (30.3%) ahead of Knight Frank (28.3%).
Keatons ranks second in general East London (35.2%) but eighth in the cross-market seller analysis (13.9%).
Lettings and management capability appears in 95.0% of lettings responses; reputation in 83.1%.
Savills leads seller outcomes and the supplementary Islington analysis, but ranks third in general East London.
The report identifies reputation, local expertise, marketing reach, valuation knowledge and customer service as recurring themes.
The research used controlled prompts designed to reflect commercially relevant questions asked by London sellers and landlords.
The prompts were run repeatedly across ChatGPT, Gemini and Google AI Overview. They were presented without wording intended to favour or steer the models toward a particular agency.
The analysis followed a two-stage process. First, every agency appearance within each complete response was detected and classified. Second, rankings were calculated using only appearances classified as genuine recommendations or shortlist inclusions.
An agency could receive a maximum of one recommendation appearance per response, regardless of repetition.
Because prompts were monitored for different periods, rankings use equal-prompt-weighted recommendation share. Each prompt therefore contributes equally to its category.
The methodology measures AI recommendation visibility rather than factual market leadership or independently verified business performance.
The study measures AI responses during the June–July 2026 collection period.
The models can change their answers because of updates, retrieval changes, new source material and prompt variation.
The research does not verify actual sales volume, price achievement, fees, customer satisfaction or property-management performance.
Claude is not present in the current London exports.
The Docklands category contains fewer responses than Central London and Islington and should be interpreted accordingly.