Rightmove's real competitor isn't OnTheMarket. It's whatever you asked ChatGPT last week.

A friend house-hunting in Leeds sent me a screenshot last week: she’d pasted three Rightmove listings into ChatGPT and asked it to tell her which was overpriced and what to check before viewing. Not a portal search. Not a saved-search email. A chat window, doing the comparing.

That’s the whole post, really. Everything below is just me working out why it matters.

What a portal actually sells. Not listings — agents own those. Not photos, not floorplans, those are the agent’s too. A portal sells aggregation: being the one place enough buyers search that no agent can afford to be absent from it. That’s the entire moat, and it’s a real one — trade press puts Rightmove’s take at around 7.2% of an average agent’s monthly commission, up to 13.5% in Glasgow, and it holds because buyers keep starting there, not because agents love paying it. It’s a search-habit tax, collected as long as the habit stays put.

The habit is the thing that’s moving. My friend didn’t stop trusting Rightmove’s listings — she just stopped doing the comparing on Rightmove’s page. That step, “is this actually a good deal, what am I missing,” used to require either expertise or laboriously opening six tabs. Now it requires pasting a URL into a chat window. The portal still hosts the data. It’s no longer where the thinking happens, and the thinking is the part a buyer actually needed help with.

This isn’t a fringe worry I’m inventing to justify a side project — Rightmove is visibly racing against it. Its own H1 2026 results describe an AI-enabled valuation tool generating roughly 50% more unique valuation leads, an “Ask Rightmove” conversational search feature aimed squarely at keeping that comparison step on-platform, and 46 AI initiatives in development — about 40% more technology releases than the same period a year earlier. A company doesn’t ship that much AI, that fast, defending a moat it thinks is safe. Source: Property Industry Eye, 31 July 2026.

And to be fair to them — it’s working, so far. Agent retention hit a 10-year high (96%) in H1 2026, revenue rose 7% to £225.8m, and that’s with a live £1.5bn lawsuit and a fresh round of fee-hike open letters from angry independents. If you’re waiting for the headline “agents flee Rightmove,” that headline doesn’t exist yet, and I’m not going to pretend it does. The habit is eroding, not collapsing.

Here’s the part that actually made me sit up, though. An “Ask Rightmove”-style feature answers questions using Rightmove’s own closed dataset — it’s a smarter front door on the same building. But my friend’s ChatGPT window doesn’t work for Rightmove. It works for her. And it’s guessing, because nothing feeds it a straight answer to “is this listed anywhere else, has the price moved, is this the same flat that fell through in March.” When the intermediary between a buyer and a listing stops being a portal’s search page and starts being a general-purpose assistant, the thing worth owning stops being the front door and starts being the answer the assistant gives when it’s asked — and that’s up for grabs in a way the front door never was, because a portal that legally can’t let an agent link off-platform (Rightmove’s own Technical Guidelines ban it outright — covered here) has no reason to build the honest, cross-source version of that answer itself.

So that’s the direction I’ve actually started sketching for Rigove, honestly, past the “and then a miracle happens” stage but nowhere near shipped: a structured feed — plausibly an MCP server, since that’s now a solved, cheap pattern to stand up — that answers exactly the questions my friend’s ChatGPT window was guessing at. Where else is this listed. Is it stale. What’s the price history. It doesn’t require anyone to find rigove.co.uk, which has been Rigove’s actual distribution problem since day one — if an assistant already queries it, a visitor never has to. I want to be clear this is a bet, not a result: the underlying data is largely there already because of what Rigove’s link map already collects, but nothing about “AI keeps intermediating property search” is proven, and I could be wrong about the timeline or the whole shape of it.

I didn’t build any of that yet. What I noticed is smaller and more useful than a prediction: the step buyers actually need help with quietly moved off the portal’s own page, months before anyone built a product for where it moved to. Whoever answers honestly there doesn’t need 80% of search traffic. They just need to be the thing that gets asked.

— Ed

Rigove is an independent tool and is not affiliated with, endorsed by, or connected to Rightmove.