UK agents report almost no AI-driven enquiries today, while consumer adoption climbs quickly and the portals have already moved inside the assistants. The available evidence says off-site brand mentions predict AI citation around three times more strongly than backlinks, and that schema markup on its own produces no measurable lift. For estate agents, the sources that matter are sector-specific, and most generic AI search advice does not map onto them.
Only 7% of UK estate agents believe AI-powered search is currently improving the visibility of their listings. In the same survey, 88% said that property businesses failing to adopt AI search risk falling behind their competitors.
Both of those numbers are probably right, and the space between them is where the opportunity sits. Nobody is losing instructions to ChatGPT this quarter. The agencies building their position now are doing it because the thing that determines whether an assistant names you takes eighteen months to accumulate and cannot be bought in a campaign.
What follows is what the evidence actually supports, what it does not, and which parts of it apply specifically to an estate agency rather than to a software company with a blog.
Are buyers and sellers actually using AI to find estate agents?
Not yet, on the evidence agents report. A survey of 622 UK estate agents carried out in February 2026 found that 22% had seen a buyer reference artificial intelligence when explaining how they found a property. Consumer-side research points the other way, and both findings can hold at once.
The GetAgent research, conducted by ProperPR via Find Out Now on 26 February 2026, found:
- 22% of agents had seen buyers reference AI when explaining how they found a property
- 17% reported sellers asking whether AI tools were being used to market their home
- 7% believed AI search was improving listing visibility, with 93% seeing no impact
- 84% expected AI search to become a primary way buyers look for property
- 88% believed non-adopters would fall behind
Set against that, BrightLocal’s Local Consumer Review Survey 2026 found the proportion of consumers using AI tools to find local business recommendations rose from 6% to 45% in a year, while Google’s share of local discovery fell from 83% to 71%. That study surveyed 1,002 US adults, so it describes a direction of travel rather than the UK property market specifically.
The two sets of figures reconcile more easily than they look. Agent-reported observation is a lagging indicator. A vendor who shortlisted three agencies through an assistant before booking a valuation will tell you they found you on Google, or will not mention it at all. Nobody narrates their research process at the door.
Why AI answers about property keep pointing back to the portals
Because the portals are now inside the assistants. Rightmove became the first UK property portal to launch an app in ChatGPT in March 2026, accessed by addressing “@Rightmove” in a prompt. Zoopla now runs one too.
The mechanism matters more than the announcement. Assistants favour sources that offer breadth and independent corroboration, and a national portal has both by definition. It carries every agency’s stock, it is referenced across the whole web, and it resolves cleanly as an entity. An individual branch in Guildford does none of those things at the same scale.
So when someone asks an assistant about property in a given town, the portal is increasingly the interface rather than the competitor. Rightmove holds roughly 80% of UK residential property search traffic, and that structural position now extends into a channel where 68% of Google searches ended without a click between January and April 2026.
The commercial risk is not that agents disappear. The pattern across every portal launch so far is that AI takes discovery and the portal keeps conversion. The risk is disintermediation at the consideration stage, meaning the period between a homeowner deciding they might sell and picking up the phone to book valuations. That stage is precisely where an agency’s differentiation normally does its work, and it is the part of the journey moving furthest out of your control.
What actually makes an AI name your agency
Being talked about, more than being linked to. Ahrefs analysed 75,000 brands and found that branded web mentions correlate with AI Overview visibility at 0.664, against 0.218 for backlinks. The three strongest signals in the study were all off-site brand signals rather than link signals.
A branded web mention is any reference to a business by name on a third-party page, whether or not it carries a link. The correlation ordering ran:
- Branded web mentions: 0.664
- Branded anchor text: 0.527
- Brand search volume: 0.392
- Backlinks: 0.218
The study also found a distribution effect worth understanding. Brands in the top quartile for web mentions earned up to ten times more AI Overview mentions than the quartile immediately below them. Visibility does not scale smoothly with effort; it compounds once you clear a threshold of third-party discussion.
Ahrefs are explicit that these are correlations rather than proof of causation, and the caveat deserves repeating rather than burying. Well-known brands plausibly earn mentions and citations for the same underlying reason. What the data supports is a reallocation of where off-page effort goes, not a claim that publishing your name in more places mechanically triggers recommendations.
The schema advice everyone repeats is not supported by the evidence
Schema markup on its own does not appear to increase AI citations. Ahrefs tracked 1,885 pages that added JSON-LD structured data between August 2025 and March 2026 against matched control pages, and found no meaningful uplift in citations across Google AI Overviews, Google AI Mode or ChatGPT. Google’s own guidance states that its AI features require no special schema and no llms.txt file.
Schema markup is structured code that labels a page’s content for machines, describing what a business is, where it operates and what a page contains. It remains worth implementing. It supports rich result eligibility, it makes entity relationships explicit, and it should always match the content visible on the page. What it is not is the lever it has been sold as.
This matters because the property trade press has spent 2026 calling schema essential for AI visibility, usually without naming a single schema type. Meanwhile the 2024 Princeton and Georgia Tech research that gets cited constantly in these discussions never tested schema at all. It tested in-content tactics, and found that adding statistics to a page lifted visibility by 41%.
The honest position is that a lot of confident AI search advice is being sold without evidence behind it, and marketing directors are being asked to fund it. Ask anyone recommending a tactic what study supports it. The answers are informative.
The sources that actually describe estate agents
For an estate agency, earned media is not Reddit and YouTube. It is the agent comparison platforms, the professional registers, the local press and the portal profiles. Those are the independent sources an assistant reads when someone asks who the best agent in a town is, and they are the ones generic AI search advice never mentions.
The map for a UK agency looks like this:
- Agent comparison platforms including GetAgent, AllAgents and EstateAgent4Me
- Propertymark membership, and the Property Ombudsman or Property Redress Scheme register
- Client Money Protection scheme membership for lettings
- Google Business Profile, maintained at branch level rather than group level
- Local and trade press coverage
- Portal agent profiles, which serve as verification that you are actively trading
For multi-branch agencies there is a second problem underneath this one. Descriptions diverge. The website says residential sales and lettings, LinkedIn says property services, the directory entry says estate agency, and the local trading name differs from the group brand. Every divergence reduces the confidence with which a model can resolve who you are, and models that cannot resolve you confidently name someone else. The same branch-level inconsistency that suppresses local rankings suppresses AI mentions, for closely related reasons.
None of this is separate from the organic work. Roar’s programme with Hamptons International started by resolving indexing problems that had hidden most of the site, taking indexed pages from 12,501 to over 21,000, and SEO now drives 55% of all website conversions. With Airsat, a single-branch independent in Bristol, average mobile rankings moved from position 22 to position 8 and the site produced 100 enquiries in 90 days, built on depth in niches larger competitors had left empty. Strong organic ranking remains the best available predictor of AI citation, which makes SEO for estate agents the foundation rather than the alternative.
How to measure whether any of this is working
Track appearance rather than traffic. AI referrals under-report badly in analytics, so the usable measure is how often your agency is named or cited across the searches happening in your branch areas, scored on the same basis every month.
Roar weights this in three ways. A full citation, where an agency is both named in the AI answer and linked as a source, scores 1.0. A name-only mention or a link-only citation each score 0.5. The scores are totalled to give an AI Visibility Score, calculated from searches run across more than 4,000 UK locations every month, tracked separately across organic listings, paid results, the map pack and AI Overviews, with four years of history behind it.
Review cadence should differ by channel. Monthly is right for organic. Quarterly is right for AI, because AI Overview content changes enough month to month that a single snapshot tells you very little about your actual position.
If you want a baseline before committing to anything, Search Radar scores any website’s visibility across Google and the AI assistants, and our benchmarking reports break the same data down to branch level.
Where this leaves you
Adoption is genuinely early, which is exactly why the advice market is currently full of confident claims with nothing underneath them. The agencies that move now are not chasing enquiries that exist today. They are accumulating the third-party footprint, entity consistency and local depth that decide whether an assistant names them in 2027, none of which can be assembled in a quarter once the demand arrives.
If you want to know where your agency currently stands and what would actually move it, our generative engine optimisation work starts with the visibility audit rather than the sales pitch.
Frequently asked questions
Does being on Rightmove and Zoopla help my AI visibility?
Yes, as independent verification that your agency is actively trading and listing real properties. Portal presence alone will not get you named, because assistants synthesise across multiple sources. Your own website content, professional memberships and review profile need to agree with the picture the portals present.
Should estate agents add schema markup for AI search?
Add it for rich result eligibility and clean machine readability, not because it will earn citations. The available causal evidence shows no measurable citation uplift from schema alone, and Google states its AI features require no special markup.
How often should we review AI visibility?
Quarterly. AI Overview content changes frequently enough that a single month’s data is unreliable as a measure of position. Organic visibility is worth reviewing monthly alongside it.
Will AI replace property portals?
Not on current evidence. The consistent pattern across portal launches inside ChatGPT so far is that AI takes the discovery stage and the portal keeps conversion, which reshapes the funnel rather than removing it.
