Public sector technology

Around 28% of central government's technology estate is classified as legacy. That figure comes from the State of Digital Government Review, published on 21 January 2025, and it is quoted far more often than the range printed beside it, which runs from 10% to 60% depending on the organisation. The range is the useful number. It says that legacy is not one national condition waiting for one national remedy. It is several hundred local situations, each with a different supplier, a different set of undocumented dependencies, and a different year in which the person who understood the system left.

An organisation at 55% is not running late on a programme. It is running a different business from one at 12%, and it should be reported as such.

The same review recorded that legacy systems counted in the annual assessment rose by 26% on the previous year, and that 22% of them were red-rated. Those are the government's own numbers about its own estate, which is the sort of admission that almost no private buyer of technology ever makes in public. It is also the reason this section carries the same weight here as the commercial ones.

The rules changed in February 2025. The market has not, yet.

The Procurement Act 2023 came into force on 24 February 2025, replacing the regime that had governed public buying for a generation. The claim made for it is transparency across the whole life of a contract rather than at the moment of award, which in principle lets an outsider see not only who won but what happened afterwards.

Whether it changes who wins is an empirical question and the evidence will arrive slowly, in notices, over several years. Anyone telling you now what the Act has done to supplier concentration is guessing. We would rather publish that finding in 2029 with the notices counted than assert it in 2026 because it would make a better headline.

What can be said already is that the transparency is only worth something if somebody reads it. Award notices, contract change notices and performance information are published into a system that almost nobody outside procurement teams and bid writers ever opens.

Local government is where most of the systems actually are

Central departments attract the coverage because they publish the most and because their failures are debated in Parliament. The systems that decide whether a housing repair gets raised, whether a care package is paid, or whether a bin round runs at all sit in councils, and a large share of them come from a small number of suppliers whose products were designed for a different decade of computing. Councils publish contract notices, spending above a set threshold and audited accounts. The picture can be assembled from public records. It is simply tedious to assemble, so it usually is not, and the result is that the part of the public sector closest to the public is the part least reported. Our coverage of that sits at councils run on systems nobody maintains.

The NHS is a third case again, and it is routinely miscategorised as part of central government. A trust is an employer, a landlord, a laboratory operator and a buyer of clinical systems, with its own board and its own capital constraints, and the money labelled national digital investment reaches it through mechanisms that are worth tracing rather than assuming. That tracing is the subject of following the NHS technology fund to the ward.

Automation in government gets its own treatment in our intelligent automation guide, because the argument there is about what a running service costs in its second year rather than what a pilot proved in its first.

A case study is written by the party with most to gain

Supplier case studies about public sector work are not reprinted here, and the reason is not squeamishness about marketing. It is that a case study almost never travels with the three documents that would let a reader judge it: the contract value, the variation notices, and whatever assessment followed. Those three exist, usually in public, and they frequently tell a different story from the one on the supplier's website. Where a claim about a public service can be tested against an audit report, a committee session, a published notice or a response to a freedom of information request, it gets tested against those. Where it cannot, the piece says who would have to publish what to settle the question, which is more useful to a reader than a confident sentence.

The reason to read public sector technology closely is not sentiment about public service. It is that public bodies are the only large technology buyers in Britain compelled to show their working, which makes them the best available evidence about a market that otherwise reports itself. What a vendor will not tell you about its renewal behaviour, its implementation overruns or its exit terms is frequently sitting in a council's committee papers.