Following the NHS technology fund to the ward

The Spending Review of June 2025 committed the government to invest up to £10 billion in NHS technology and digital transformation by 2028-29, which it described as an almost 50% increase from 2025-26. Two days before the Autumn Budget, which the Chancellor presented to Parliament on 26 November 2025, the Treasury trailed a further £300 million of new capital investment for NHS technology, to fund digital tools that automate administrative tasks, speed access to patient information and improve coordination. The Budget document itself records the £300 million as capital investment in the NHS without breaking out the technology purpose, which is the first small illustration of the problem this piece is about.

The same Spending Review states the return expected: funding at this level will enable the NHS to deliver 2% productivity growth each year, unlocking £17 billion of savings over three years.

Those three numbers form a chain of reasoning that appears in almost every ministerial statement about health technology. Money goes in, productivity comes out, savings follow. This piece is about whether the middle link can be observed.

The productivity numbers are real and they are good

Start by conceding the point, because the figures are stronger than sceptics usually allow.

NHS England’s productivity plan update of 4 February 2026 reports a 2.7% improvement in 2024-25 and 2.6% acute productivity growth in the first half of 2025-26, rising to as much as 3.4% overall when non-acute providers are included. Against a long-run average of around 0.6% a year, that is a substantial change in a service where productivity growth has been the single hardest thing to move.

The operating detail behind it is specific. Between April and November 2025, compared with the same period a year earlier, non-elective bed days fell 1.8%, elective ordinary length of stay fell 2.0%, agency spending fell 42% and bank spending fell 7%. Theatre utilisation rose 1.7 percentage points, and that one is a December 2025 figure rather than a reading over the same eight months, which is easy to miss because it sits in the same table.

Look at that list again. Four of the five are workforce and flow decisions. The one that most obviously moved the total, a 42% cut in agency spend, is a procurement and rostering discipline, not a piece of software.

What NHS England says about attribution

The methodology is where the chain breaks, and NHS England is not hiding it.

Its published account of how NHS productivity is measured describes the exercise as measuring how well the service turns a volume of inputs such as staff, drugs and medical equipment into a volume of outputs. It identifies the drivers it can see behind changes in the figure: case complexity, staffing turnover, industrial action, and depreciation from capital investment. It notes that the analysis has not yet been able to include all acute activity, diagnostics among it.

Technology does not appear in that causal list. The measurement was not designed to isolate it, and it does not.

That is a reasonable position for a statistician and an awkward one for a business case. If the productivity index cannot separate the contribution of an electronic patient record from the contribution of a shorter length of stay driven by a discharge coordinator, then no amount of aggregate improvement demonstrates that the technology spending worked. It demonstrates that the service improved while the technology spending was happening, which is a different claim and one that would also be consistent with the money being wasted.

Where technology does show up specifically

There are places where the connection is drawn tightly enough to be useful, and they are the parts of the plan worth reading.

The clearest is the NHS App. NHS England reports that providers fully connected to all the core services of the app see a 2.5 to 3% improvement in their waiting lists. That is a stated effect, attached to a stated condition, at provider level. It can be checked as more trusts connect, and it fails if the improvement does not follow the connection.

The scale it acts on is also published: NHS trusts delivered around 53 million outpatient follow-up appointments in 2024-25, costing £9.0 billion. Any percentage of that pool is a large number, which is precisely why outpatient flow is where digital tools are being pointed. The Wayfinder programme, which pushes appointment information into the app, is reported to have exceeded its benefits target for the year with major reductions in the rate of patients who do not attend.

A missed appointment is the cleanest thing digital health can affect. The patient does not turn up because they did not know, could not change it, or forgot. A notification and a rebooking button address all three, the effect is measurable within weeks, and nothing about clinical practice has to change. It is not glamorous and it is the strongest evidence in the whole file.

The money does not arrive as a fund

Part of the tracing problem is definitional, and it is worth being precise about it because the phrase technology fund does a lot of unexamined work in this debate.

There is no single account from which NHS technology money is drawn and into which a ward can look. The Spending Review sets a departmental envelope. Some of it flows as national programme spending, buying platforms and services centrally. Some flows to integrated care boards and on to providers. Some arrives as capital allocations that a trust board then prioritises against estates, equipment and backlog maintenance, and a trust facing a failing roof and an ageing scanner does not always choose the software.

Each of those routes is legitimate and each is accounted for differently. The consequence is that the up to £10 billion is a plan for a departmental total, not a traceable transfer, and any attempt to follow it to a specific ward runs into three different accounting boundaries before it gets there.

That is not concealment. It is how health capital has always worked. It does mean that the sentence money went into NHS technology and productivity rose is doing considerably less work than it appears to.

What is not traceable

Now the honest gaps, stated plainly rather than hedged.

There is no published route from the up to £10 billion to a ward. The Spending Review sets the envelope. NHS England’s plan describes programmes. Individual trusts publish capital plans and annual accounts. Nothing links the three into a line that says this sum bought this capability in this organisation and this is what changed afterwards. Assembling it would mean reading several hundred sets of trust accounts against a national envelope defined on a different basis, which is why nobody does it.

There is no published counterfactual. Productivity is rising during a period of unusually intense national attention on productivity, including the agency spend reduction, the operational planning targets and the ending of industrial action that suppressed earlier figures. Some of the improvement is technology. Nobody has published a figure for how much, and anyone who quotes one has produced it themselves.

And the ratio of capital to running cost is not stated. A record system, an app integration and an automated triage tool all carry annual costs after the capital is spent: licences, integration maintenance, support, and clinical safety work when the supplier ships a change. Those costs land on revenue budgets in trusts, not on the national capital line, and they are the reason a business case that clears on capital can fail in year three. That arithmetic is worked through in what automation costs to run.

The commitment that will test all of this

The 10 Year Health Plan for England, published on 3 July 2025, organises itself around three shifts, one of which is analogue to digital. The concrete expression of that shift is the single patient record. The government’s announcement of 15 May 2026 states that all NHS providers, hospitals and GPs among them, will have to share data so that clinicians across England can securely see a patient’s full medical history, with improved access for clinicians as early as 2027 in specialties including maternity and frailty care.

That is the hardest thing on the list, because it is not a purchase. A shared record is an agreement between several hundred organisations about what a field means, who may write to it, and who carries the consequence when two entries disagree. Previous national attempts in the NHS ran aground on that rather than on the engineering.

If it lands, the productivity argument gets easier to make, because a record that follows a patient removes duplicated tests, repeated history-taking and the administrative work of chasing information between organisations. Those are measurable at the point they stop happening, which is more than can be said for most of what is claimed in this field. If it slips, the reason will be worth reporting in considerably more detail than the announcement was.

What would settle it

Three publications would convert this from an argument into a measurement, and all three are within the gift of bodies that already collect the data.

Trust-level connection status against trust-level waiting list movement. NHS England has both. Publishing them together would let anyone test the 2.5 to 3% claim rather than accept it, and would make the claim more credible if it holds.

Total cost of ownership alongside capital allocation. If a trust receives capital for a system, publishing the recurring cost it takes on would show whether the national investment is transferring a liability to local revenue budgets. Boards already approve this figure internally.

And a productivity measure that separates technology-attributable activity, or an explicit statement that it cannot. The second would be more useful than the current position, in which the index is quoted in support of technology spending by people who have not read the methodology and would not find technology in it if they had.

The implication

None of this argues the money should not have been spent. An estate that spent a decade buying almost nothing needed a period of buying, and the case for record systems, app connectivity and automated administration does not depend on a productivity index.

It argues something narrower and more awkward. The health service is currently running the largest technology investment in its recent history while measuring its own performance with an instrument that its own documentation says does not isolate the effect of technology. That is a governance gap, not a technical one, and it will be filled either by the bodies that hold the data or by whoever wants to claim the credit.

The rest of this coverage sits under healthcare technology and public sector technology.