Rightmove CEO Dismisses AI Threat to Property Portal Business Model
The chief executive of Rightmove Plc says artificial intelligence presents more opportunity than existential risk to property listing platforms, pushing back against concerns that AI-powered search could disintermediate the online real estate marketplace.
The comments come as finance leaders across digital platforms grapple with whether large language models will eventually route users directly to listings—bypassing the aggregators that have dominated online real estate for two decades. For CFOs at marketplace businesses, the question isn't academic: their entire revenue model depends on being the destination, not a data source for someone else's AI.
Rightmove, which operates the UK's largest property portal, has built its business on a straightforward proposition: estate agents pay to list properties where buyers are already looking. It's a high-margin model that has made the company a cash machine. But that model assumes buyers start their search on Rightmove's website, not by asking ChatGPT or Google's AI where to find a three-bedroom flat in Manchester.
The CEO's confidence that "peers will survive" suggests he sees AI as a feature enhancement rather than a structural threat. (Translation: we'll add AI search to our existing platform, not get replaced by it.) That's the optimistic read. The pessimistic read is that every marketplace CEO says this until the traffic numbers start declining.
Here's the thing finance leaders are actually worried about: AI doesn't need to replace Rightmove to damage its economics. It just needs to change user behavior at the margins. If 15% of property searches shift to AI-first interfaces that pull data from multiple sources, that's 15% fewer page views, fewer ad impressions, and—eventually—less pricing power with estate agents. The unit economics get worse even if the business doesn't collapse.
The "AI opportunities" the CEO references likely include the standard playbook: better search relevance, automated property descriptions, predictive pricing tools for agents. These are table stakes. Every property portal will have them within 18 months. The question is whether they're enough to maintain the moat when users can increasingly bypass the portal entirely.
What makes this particularly interesting for finance operators is the disclosure problem. How do you quantify "AI risk" in a 10-K when the threat is behavioral shift, not direct competition? You can't point to a specific competitor stealing share. You're fighting the possibility that search behavior itself is changing. That's much harder to model—and much easier for management to dismiss as speculative.
The survival question isn't whether Rightmove specifically makes it (they probably do—they're the incumbent with the data and the agent relationships). It's whether the margins survive. If AI makes property search more efficient, that efficiency might accrue to users and agents, not to the platform. That's the CFO nightmare: revenue holds up, but pricing power erodes, and suddenly you're in a lower-multiple business.
For now, the CEO's confidence is a data point. Whether it's justified depends on something finance teams are watching closely: where the traffic actually comes from in 12 months. If direct navigation holds steady, he's right. If it doesn't, the "opportunities" conversation gets a lot more complicated.


















Responses (0 )