Your conveyancer got a machine. You got a bill that is the same and a wait that is longer

Somebody bought a house this month and paid about £2,438 for the legal work. A year ago they would have paid about £2,437.

In cash that is a rise of one pound. Set against inflation running at 2.6 per cent it is a real-terms fall, and the man whose company publishes the index says so himself. Rob Houghton of reallymoving has now said it across four consecutive quarterly releases: growth is flat, adjusted for inflation costs have fallen, and conveyancers are earning less per transaction than they were a year ago.

So the customer got something. Not much, and nobody told them, but something.

Over the same period eight in ten conveyancing firms started using AI to support their fee earners, double the proportion a year earlier. Nearly half say they are investing specifically in AI to make their workflows faster.

And the transaction got slower. The average time from instruction to completion reached 123 days in 2025, up 18 per cent since 2019.

Those two facts sit next to each other and I am not going to draw a line between them. Time to completion covers local authority searches, lender underwriting, management company replies and Land Registry processing, and none of that sits on a conveyancer's desk. The industry's own advice to firms says as much: the solicitor part gets faster, the everyone else part does not.

But hold the shape of it. The technology arrived almost everywhere. The price fell slightly. The firms say they are poorer per job. The buyer waits longer than in 2019. Somewhere in there is a saving, and I cannot find anybody who admits to having it.

Then we went looking for the foundation the whole thing rests on: the evidence that this technology made the work faster.

There is a figure every conveyancing partner in the country now quotes. Five hours a week per lawyer, saved on title checks at a firm called Talbots Law. We tried to firm it to Talbots. It does not come from Talbots. It comes from a case study published by Orbital, the AI platform Talbots bought, republished by the trade press with a line at the bottom saying the article was submitted by Orbital. The underlying quote is one man saying that if you do a lot of title checking it could easily save you five plus hours.

So we looked for another. Eighty-four per cent of solicitors saving up to six hours a week: published by LEAP, which sells practice management software to law firms. The collapse in paralegal hiring: a survey of GenieAI's customers, reported with a quote from GenieAI's chief executive.

Three claims. Three suppliers. We have not found a single independent measurement of what AI has done to the speed of legal work in this country.

I am not saying nobody has measured it. I am saying we looked and did not find it, and those are different sentences. The second is a claim about the world. The first is a claim about us, and it is the only one I can stand behind.

This matters well beyond conveyancing. This is a service with a published price series going back to 2018, because a regulator required one. It is among the very few. If the evidence base is this thin where the data is unusually good, consider the sectors where nobody publishes anything at all.

So we are writing to people. To the Solicitors Regulation Authority, the Law Society and the Legal Services Board, to ask whether any of them holds an independent measure. And to Orbital, LEAP and GenieAI, whose numbers we have called unverified in this edition, because if you say that about a company you say it to them on the same day. Every letter is logged before it is sent. When answers come back we will publish them, prominently, whether they help us or embarrass us.

That is a change in how this paper works, decided this week. We are going to ask people things now, rather than only reading what they chose to publish.

One last thing, and it is the reason this edition exists at all. We built this piece expecting to find that prices had held while the work got easier. The numbers said the opposite. The price did fall, quietly, in real terms, and we would have missed it, because we were looking for harm.

The abundance turned up. It was small, nobody announced it, and the jobs below qualified level went anyway.

— M.

This note is mine: the view, and the call to run it. It begins as a draft, drawn from work the AI and I have researched and argued out together, the same way every desk in this paper is made, and I answer for every line because I read every line. Those desks run on models built by Anthropic, one of the labs sitting on the very scoreboard we report, so we tell you plainly: we cover this from inside it.