The HMCTS court data licence now sets an explicit limit on what legal technology may do with court listing data. Published on 15 July 2026, it caps four high-risk uses, including anticipating legal decisions for a client, at a one-year Experimental licence that cannot be released to customers at all. A three-year Transactional licence covers commercial products everywhere else.
What does the licence actually cover?
The subject matter is narrower than “court data” suggests, and the narrowness matters. What the Third-Party Courts and Tribunals Data Licence grants is permission to carry out computational analysis on the lists published through the Court and Tribunal Hearings Service, in which the Crown holds database rights. Without a licence, a business scraping those lists is not merely operating in a grey area; it has no permission to copy, adapt or redistribute the underlying data.
Two licences are on offer. The Experimental licence runs for a year and is confined to proof-of-concept work, which cannot be released either to customers or to the public, so a product built under it stays in the laboratory until the position changes. The Transactional licence runs for three years and permits commercial products. Both are free on initial grant, with HMCTS reserving the right to charge in future, and applications are made through the departmental guidance published alongside the licence.
Which uses are treated as high risk?
Four categories will only ever be offered an Experimental licence, which is the operative restriction in the whole document and the reason the Law Society Gazette picked the story up on 3 August. They are: fully automated legal advice; anticipating legal decisions for a client or consumer through predictive analytics or modelling; otherwise directly informing whether a third party pursues legal action; and personal profiling, particularly where court data is combined with other datasets.
Read together, those four describe a good deal of what litigation analytics is sold as doing. A tool that tells a client how a judge has decided comparable applications, or that scores a claim’s prospects before a letter before action goes out, falls within the second or third category on any sensible reading. That does not make it unlawful, and HMCTS has not said it may never reach the market. It does mean the route to market runs through a one-year licence with no commercial release, and then through a renewal decision that nobody has yet made.
How the Governance Panel decides
Applications go to a Governance Panel, which weighs them against twelve principles covering open justice, judicial independence, protection against bias and algorithmic transparency, among others. Applicants must supply a data protection impact assessment, and, where relevant, a completed algorithmic transparency recording standard template. The application guidance sets out what each stage requires.
One feature deserves more attention than it has had. HMCTS will publish both successful and rejected applicants, with reasons, which turns a refusal into a public document about a named business and its intended product. Appeals must be brought within 20 working days of the decision. For a supplier, that combination changes the calculation: an application is not a low-cost enquiry that can be withdrawn quietly if it goes badly.
The timings are worth building into a product plan rather than discovering later. An Experimental licence lasts 12 months against a Transactional licence’s 36, so a proof of concept that takes nine months to build leaves three in which to demonstrate anything to the Panel that granted it. Nothing in the guidance promises that a renewal will follow, and the licence has been live for barely three weeks, so there is no track record of Panel decisions to reason from yet.
What this means for procurement and diligence?
If a firm buys, builds or invests in litigation analytics, the licence terms are now the first diligence question rather than a late one, because they determine what the product is permitted to do with its source data before anything else about it matters. The questions worth asking a supplier are short: which licence do you hold, when does it expire, and which of the four high-risk categories does your roadmap touch?
The wider significance is that a public body has, for the first time, written down where the line falls on predicting outcomes from court data. Every analytics product in the market has had to take a private view on that question. There is now a published one to be measured against, and the published version is more cautious than most of the marketing. It sits within the direction of travel the Master of the Rolls set out when he argued that lawyers must help shape machine-age justice rather than react to it. On the documentation side, the DPIA requirement is the same discipline firms are already meeting on data protection complaints processes, so the drafting work is not starting from nothing.
If your firm buys, builds or invests in litigation analytics, the licence terms now determine what a product can lawfully do with court listing data. Anything that predicts outcomes for a client, or tells them whether to sue, is capped at a one-year experimental licence with no route to market until it is renewed as transactional. Procurement and diligence questions should start here.