The most important fact buried in the latest records surrounding the NHS Federated Data Platform is not £330 million, £808 million, £1.08 billion or even £2.4 billion. It is a date.
April 2029.
That is when the independent evaluation of the Federated Data Platform commissioned by NHS England and led by experts from Imperial College London is due to produce its final report. The evaluation has been designed to examine the platform’s impact, implementation, unintended consequences and value for money, while attempting to account for precisely the problems that now hang over some of NHS England’s most widely promoted evidence: seasonal effects, concurrent NHS initiatives, changes in patient case mix, differences between hospitals and the difficulty of establishing causation from before-and-after comparisons. Imperial’s newly published evaluation strategy explicitly states that an observational design cannot establish definitive causality and sets out more sophisticated methods intended to get closer to a reliable answer. Its final report, however, will not arrive until April 2029.
The decision over the current supplier comes considerably sooner.
NHS England awarded the Federated Data Platform and Associated Services contract in November 2023 to a consortium led by Palantir Technologies UK, alongside Accenture, PwC, Carnall Farrar and NECS. The platform itself is built on Palantir’s Foundry software. NHS England describes the arrangement as worth up to £330 million over seven years, but the seven years were never committed in one block: the initial term runs for three years, followed by potential extensions of two years and then two further one-year periods. The initial term ends in March 2027. NHS England’s own contract explainer confirms that structure, while its FDP frequently asked questions confirms the £330 million maximum value.
That creates an uncomfortable chronology. The NHS must determine whether to continue the Palantir-led arrangement while the most substantial independent investigation into whether FDP is delivering what has been claimed for it remains years from completion.
The importance of that gap becomes clearer when the evidence already being used to justify the programme is examined.
The review that is not the evaluation
In June, the Health and Social Care Committee put the timing problem directly to NHS England and the Department of Health and Social Care.
During a 16 June evidence session, FDP programme director Ayub Bhayat confirmed that notice would have to be given by December if the agreement was not being extended. More importantly, he distinguished the assessment taking place now from the Imperial evaluation.
The analysis informing the immediate contract decision, he told MPs, was “not the Imperial College evaluation”. It was a commercial review being conducted under the Cabinet Office contract-review process, expected to report in early autumn.
That distinction matters enormously.
Imperial has been commissioned to examine whether FDP achieves its objectives and outcomes, how it is implemented, what unintended consequences it produces and whether it represents value for money. The commercial review occurring now will consider the existing agreement in time for ministers to decide whether the Palantir-led consortium remains in place beyond March 2027. According to NHS England’s description of the extension process, that review will consider uptake, product development and benefits alongside NHS and departmental strategy, the supplier market and commercial strategy.
The government therefore does have an evaluation mechanism before making its decision. It would be wrong to suggest otherwise.
What it will not have is the completed independent national evaluation specifically commissioned to produce a more robust assessment of FDP’s impact and value for money.
The same parliamentary hearing exposed another difficulty. NHS England told MPs that the original FDP procurement had taken more than 15 months. When repeatedly asked whether a new supplier could be procured and in place by March 2027 if the current arrangement were not extended, Bhayat eventually said he did not yet have an answer and was waiting for commercial advice. NHS England’s stated intention was to reach the relevant position by early autumn rather than wait until the December deadline.
There is no evidence in that exchange that the contract has already been predetermined. Bhayat explicitly rejected the suggestion that commissioning the long-running Imperial evaluation implied an assumption that Palantir would automatically be retained. But the practical imbalance is obvious: the government must soon decide whether to remain with an incumbent platform already embedded across much of the NHS, while an alternative could require procurement, migration and implementation work for which officials were, as of June, unable to give Parliament a confident timetable.
This is not merely a Palantir problem. It is a procurement problem, an evidence problem and, increasingly, a sequencing problem.
£780 million became £2.4 billion
The history of FDP’s projected benefits makes the need for independent evaluation particularly acute.
The Department of Health and Social Care’s Accounting Officer Assessment records that the full business case estimated benefits of approximately £780 million over a seven-year appraisal period. That consisted of £304.9 million in potential cash-releasing benefits, £282.3 million in non-cash-releasing benefits and £189.6 million in societal benefits. The business case produced a stated benefit-cost ratio of 4.92.
By April 2026, the figure being presented in Parliament had become much larger.
During a Westminster Hall debate on 16 April, health minister Dr Zubir Ahmed said FDP was improving efficiency and generating savings worth up to £2.4 billion, attributing that number to “independent estimates”.
Six days later, Hansard published a formal ministerial correction.
“Independent estimates” became “our estimates.”
The rest of the £2.4 billion claim remained.
That correction is not evidence that the £2.4 billion estimate was fabricated or improper. It establishes something narrower and important: the number initially presented to Parliament as independently estimated was, according to the corrected official record, an estimate belonging to the NHS or government itself.
Two months later the number was still being used.
Giving evidence to Parliament on 16 June, NHS England national medical director Dr Claire Price-Forbes described FDP as having a ten-year programme benefits profile of £2.4 billion, including £646 million in cash-releasing benefits, and said £217 million in monetary benefits had already been released. She also pointed to widespread adoption and improvements associated with FDP products. The committee transcript records those claims in full.
Then came the latest government major-project data.
£1.084 billion of cost. £808 million of forecast benefits.
The 2025/26 National Infrastructure and Service Transformation Authority return records a departmentally agreed whole-life cost of £1,084.31 million for the wider Federated Data Platform programme.
It records departmentally agreed forecast benefits of £808.28 million.
The government states that whole-life cost increased because of delivery at scale, particularly during 2025/26 and 2026/27. It simultaneously says forecast benefits decreased following refinements to the product roadmap and rollout plans, improved data, more detailed delivery assumptions and, significantly, a stronger understanding of product performance, usage and adoption. FDP nevertheless retains a Green delivery-confidence rating, and the same return says 129 acute trusts were actively using FDP products by April 2026.
There is an obvious temptation to place £1.084 billion of programme cost beside £808.28 million of forecast benefits, subtract one from the other and announce that FDP is forecast to lose hundreds of millions of pounds.
That would be bad journalism.
Nor can the £2.4 billion figure cited to Parliament simply be described as having “collapsed” to £808 million.
The figures appearing across the public record are not clearly demonstrated to be like-for-like. The original £780 million business-case estimate covered a seven-year appraisal period; the £2.4 billion figure given to Parliament was explicitly described in June as a ten-year programme profile; NISTA’s current £808.28 million figure sits within a different major-project reporting framework. Until NHS England publishes a clear reconciliation of the assumptions, time horizons, price bases and benefit categories behind those figures, treating them as interchangeable would create precisely the kind of statistical confusion this investigation is about.
The legitimate question is more difficult for the programme.
How did the public presentation of FDP benefits move through figures of roughly £780 million and £2.4 billion while the current government major-project return now reports £808.28 million, and how are those estimates supposed to relate to one another?
The answer may be perfectly defensible.
It simply needs to be made intelligible.
The numbers underneath the sales pitch
The problem does not end with financial forecasts.
NHS England’s current FDP benefits page reports that 139 trusts were live on the platform by the end of May 2026 and 137 were reporting benefits. It says the Inpatient Care Coordination Solution was associated with 111,589 additional patients undergoing theatre procedures compared with periods before FDP use. It reports reductions in average delayed days after OPTICA deployment of 13.58% for patients staying more than seven days, 13.82% for those over 14 days and 15.25% for those staying 21 days or longer.
These are substantial numbers. They have helped form the public case for FDP.
On the same page, however, NHS England now explains something essential about how part of this evidence should be read.
The performance calculations are observational before-and-after comparisons. Because other variables have not been controlled for, NHS England says it cannot draw conclusions about cause and effect.
That qualification did not always appear so prominently.
The Office for Statistics Regulation records that NHS England added an explicit causality sentence to its methods information on 6 June 2026. OSR said the change followed an expansion of methodological information and came partly after numerous Freedom of Information requests seeking details of how FDP-derived data had been analysed. Following its casework investigation, the regulator secured commitments from NHS England to place the caveat on its main benefits page, label before-and-after observations appropriately in public communications, make methodological changes clearer, publish more trust-level information and clarify that FDP case studies displayed on the website are written by organisations using FDP and are not authored or verified by NHS England.
Again, none of that establishes that the headline figures are false.
It establishes that the evidential language around them required strengthening.
That distinction matters because observational evidence can describe what happened after a technology was introduced without establishing what would have happened in its absence. Hospitals do not exist inside sealed laboratories. Staffing changes. Winter arrives. Waiting-list initiatives begin. Funding changes. Patient populations vary. Other digital systems are introduced. A before-and-after improvement can be real while still having multiple causes.
This is precisely the problem an independent evaluation is supposed to solve.
Then another analysis looked at OPTICA
The most serious challenge to the FDP benefits narrative concerns OPTICA, the product intended to support patient discharge.
The Health Foundation conducted its own analysis using publicly available data and information about participating trusts supplied by NHS England. In a letter subsequently released through the Office for Statistics Regulation, Health Foundation chief executive Dr Jennifer Dixon said the organisation found no meaningful difference in delayed-discharge performance between trusts that had adopted OPTICA and those that had not. It found no noticeable improvement in the trajectory of discharge performance after implementation and concluded that it could find “no evidence of measurable improvements” in discharge performance among trusts using the tool. The Health Foundation’s findings are now published through OSR’s correspondence record.
That sounds devastating until the methodological dispute is examined.
The Health Foundation’s analysis and NHS England’s analysis did not study identical cohorts. As the statistics regulator subsequently explained, the Health Foundation looked at all patients, including those staying fewer than seven days, whereas NHS England restricted its relevant analysis to patients staying more than seven days, consistent with the intended focus of OPTICA on more complex discharges.
That is an important defence of NHS England’s methodology.
It is not, however, a resolution of the dispute.
OSR said the contrasting information was creating a “confused picture” for patients and the public. NHS England agreed to re-engage with the Health Foundation, make OPTICA a priority within the national evaluation, apply causality caveats consistently to communications around discharge-delay claims and clarify the provenance of case studies. Health Foundation analysts are due to meet NHS statisticians in September.
The independent regulator has therefore not ruled that FDP failed, and it has not ruled that NHS England’s calculations are invalid.
It has effectively said that the competing pictures need to be reconciled.
That is a very different proposition from the certainty with which headline benefits have often travelled through political and public communications.
NHS England is changing the calculation
There is another development.
On 3 August, the Financial Times reported that NHS England intends to modify the way two major FDP performance measures are calculated after internal staff raised concerns about the statistical baseline. According to emails obtained by the FT under Freedom of Information legislation, the changes concern the approximately 110,000 additional theatre procedures and the discharge-delay metric; one internal concern was that the earlier comparison period could be distorted by seasonal patterns. NHS England told the FT revised baselines would be reflected in the next quarterly publication.
That report should not be converted into a claim that NHS England knowingly published false statistics. The available evidence does not establish that.
It does establish that two headline metrics used to demonstrate FDP’s performance are now due to be recalculated after methodological concerns were raised internally.
That becomes particularly significant when read beside the methodology the independent evaluators have now chosen.
Imperial is designing the test NHS England did not yet have
Imperial’s evaluation strategy is unusually illuminating because it explains, in technical language, why causal claims around FDP are difficult.
The evaluators identify varying data completeness and definitions between trusts and over time. They intend to check datasets for completeness, internal consistency and cross-source validation before analysis, document reporting anomalies and use sensitivity analysis where necessary.
For impact assessment, they propose interrupted time-series analysis and, where feasible, controlled interrupted time-series and difference-in-differences comparisons. Those methods are intended to account for underlying trends that may have occurred irrespective of FDP.
The reason is set out plainly in the strategy: observational analysis cannot establish definitive causality, and observed changes may be affected by concurrent initiatives, secular trends, seasonal effects and changes in patient case mix. Even the more sophisticated design, Imperial cautions, cannot completely eliminate residual confounding or establish direct causality.
This is not an indictment of FDP.
It is an indictment of certainty.
If the independent researchers tasked with answering the question say that attributing changes to FDP requires controlled comparisons, pre-specified analysis, investigation of trends and careful language around causality, then spectacular before-and-after figures should never have been treated as though the attribution question were already settled.
The test is now being built.
It simply comes after the sales pitch.
And its final answer comes after the supplier decision.
There is evidence that FDP works
Any investigation that stopped there would be incomplete.
There is evidence in FDP’s favour.
NISTA currently rates the programme Green, meaning the government’s own major-project assessment considers successful delivery to appear highly likely. The programme says benefits realisation is accelerating, ten products are generally available and adoption has continued to increase. NHS England reports 139 trusts live by the end of May and 170 signed up.
There is also peer-reviewed evidence concerning one FDP product.
A 2026 BMJ Health & Care Informatics study examined the Inpatient Care Coordination Solution’s theatre-management tool at Chelsea and Westminster Hospital using interrupted time-series analysis. It reported positive post-implementation changes in measures including booked and actual theatre utilisation and bookings per session. That is real evidence supportive of the argument that FDP tools can improve NHS operations, although it remains observational evidence from a particular NHS trust rather than proof that every national benefit attributed to the wider programme was caused by FDP.
There is an important principle here.
A serious investigation should not need FDP to fail.
If Imperial ultimately concludes that the platform produces substantial benefits and delivers value for money, that finding should be reported with exactly the same seriousness as evidence pointing in the other direction.
The issue is whether the state should establish that case before making consequential long-term supplier decisions wherever practicable, or construct the definitive evidence after the infrastructure has already been scaled and the next contractual decision has passed.
£330 million is not £1.084 billion
There is one further distinction worth making because it has already been blurred elsewhere.
The £330 million figure is the maximum value of the seven-year FDP and Associated Services contract awarded to the Palantir-led consortium. It is not the entire lifetime cost of the government programme.
NISTA’s £1.084 billion figure is the whole-life cost of the wider Federated Data Platform programme.
Those are different scopes.
The programme cost did not suddenly jump from £330 million to £1.084 billion, and writing that it did would be misleading.
The distinction does, however, create another legitimate transparency question. NHS England’s own November 2023 governance minutes described an overall FDP programme contract value of £480 million, comprising £330 million for the platform arrangement and £150 million for a multi-vendor solution exchange. NISTA now reports a wider whole-life programme cost above £1 billion. The contemporaneous Check and Challenge Group minutes document the earlier £480 million structure.
The public record reviewed for this investigation does not provide a simple itemised bridge explaining every component between those supplier arrangements and the £1.084 billion whole-life programme figure.
That is not evidence that money is missing.
It is a reason to ask for the reconciliation.
The decision point
None of this occurs in a political vacuum.
On 9 July, the cross-party Health and Social Care Committee urged the government to move away from Palantir and prepare an alternative, citing concerns including public trust, efficacy and the availability of other approaches. The Committee itself highlighted NHS England’s clarification that its before-and-after metrics could not establish FDP as the cause of observed improvements.
The government does not have to accept that recommendation.
Nor does the evidence presently available justify declaring that Palantir’s technology has failed the NHS.
Something more interesting has happened.
A programme sold through striking claims of operational improvement is reaching its supplier decision point just as the evidential scaffolding beneath some of those claims is being examined more aggressively than ever before. The statistics regulator has required clearer caveats. An independent analysis has produced a materially different picture of one key product. NHS England is reportedly changing the methodology behind two prominent performance measures. Government benefit forecasts have been revised. And the independent national evaluation now being constructed explicitly recognises that seasonality, simultaneous interventions, case mix and other confounders complicate causal attribution.
That evaluation finishes in April 2029.
The Palantir-led contract’s initial term finishes in March 2027.
The commercial assessment intended to inform what happens next is happening now.
There may ultimately be a compelling case for extending the contract. FDP may prove to be one of the better digital investments the NHS has made. The benefits may survive more rigorous analysis and the government’s shifting financial estimates may prove reconcilable once their different periods and methodologies are properly explained.
But those are conclusions that evidence should produce.
They are not assumptions that evidence should be asked to validate afterwards.
The question ministers now have to answer is therefore larger than Palantir.
What standard of evidence is sufficient to renew critical national digital infrastructure when the independent evaluation designed to determine its true impact and value for money will not deliver its final answer for another two and a half years?
For a programme built around making better decisions from better data, it is an unusually important test.
And this time, the data should come first.
Support This Work
Bitcoin (BTC) - bc1qevvy4y7ph5nxhsux0j6llfjepn239r52rakgcy
Ethereum (ETH) - 0x26F16D2D4d3dE1ab332deeE9d7DECA6B90654717
XRP - r4UDiUq5U8cQv5gq2zBdimtoMVxxjqzcYH
Solana (SOL) - 22SruEARKvXAKn8TzZZ1dBogQXRYyWkuYA4KYZWa4vAS
Dogecoin (DOGE) - DBjnCWtW1r7bnTorAocE5ypsgA1h7kQ1PD
Cardano (ADA) - addr1q8u5ld73yzv0ep9vguaz2zepdlv3ldpdr23ytukpk7lt9r8ef7mazgycljz2c3e6y59jzm7er76z6x4zghevrda7k2xq43ad8d
Sources
NHS England — FDP Check and Challenge Group minutes, 24 November 2023
Hansard — ministerial correction to the £2.4bn FDP benefits claim, 22 April 2026
House of Commons — Health and Social Care Committee oral evidence, 16 June 2026
House of Commons — oral evidence on FDP’s £2.4bn ten-year benefits profile, 16 June 2026
The Health Foundation / OSR — analysis of OPTICA discharge performance
Office for Statistics Regulation — response to Health Foundation OPTICA analysis, 10 August 2026
Imperial — NHS Federated Data Platform Evaluation Strategy, August 2026
Financial Times — NHS England to revise FDP performance data after staff concerns, 3 August 2026