Green Is Not Enough
Researched by an agentic pipeline · reviewed and gated by the author
A green delivery rating is useful precisely because it is limited.
The comforting green
A programme board sees a green delivery assessment and hears a reassuring message: the commitment is under control. Milestones are credible, delivery machinery is functioning and the route ahead remains navigable. Yet none of those judgements answers a different question: whether the remaining commitment is still worth making.
That distinction is becoming operationally important in the NHS Federated Data Platform programme. The National Infrastructure and Service Transformation Authority describes a Delivery Confidence Assessment as a snapshot of the likelihood of successful delivery, not a comprehensive view of performance. Recent reporting has placed forecast whole-life cost at £1.1 billion and monetised benefits at £808 million, while the programme remains delivery-green. Those figures are not, by themselves, proof of a negative value case: their scope, timing and price bases require reconciliation, and important benefits may not be monetised. But the divergence is enough to demand a separate judgement. [S1] [S2]
The governance error is not that a green rating is wrong. It is asking the rating to answer a question it was not designed to answer.
A programme can be deliverable without the remaining commitment being the best use of money, attention or strategic freedom.
Two connected judgements
Boards need two explicit confidence judgements.
Delivery confidence asks whether the agreed objectives can be achieved to the current plan. It considers schedule, capability, dependencies, risk and execution discipline.
Value confidence asks whether the remaining commitment is still justified. It considers comparable remaining cost, causally credible benefits, strategic necessity, public or customer value, dependency, switching cost, lost alternatives and the declining ability to change course.
The two judgements are connected but not interchangeable. A programme can have high value confidence and weak delivery confidence: the objective matters, but the execution model is failing. It can also have strong delivery confidence and weak value confidence: the machinery works, but the destination, economics or available alternatives have changed.
The point is not to create a second traffic light. It is to prevent a delivery signal from silently becoming an all-purpose endorsement. A useful board view states what each judgement covers, the evidence behind it, the uncertainty that remains and the next date on which a real option can be exercised.
| Judgement | Core question | Typical evidence | Board response |
|---|---|---|---|
| Delivery confidence | Can the current commitment be delivered? | Milestones, capability, dependencies, risk exposure | Intervene in execution |
| Value confidence | Is the remaining commitment still justified? | Remaining cost, causal benefits, alternatives, dependency, reversibility | Continue, rescope, pause or exit |
Why value confidence decays differently
Delivery confidence usually changes through visible execution events: missed milestones, supplier failure, unresolved dependencies or capability gaps. Value confidence can weaken while delivery improves.
Successful adoption creates momentum. More users depend on the platform, more processes are redesigned around it and the cost of switching rises. Those developments may increase value, but they also reduce reversibility. They can make the programme easier to defend before the benefits are causally established. Sunk cost becomes a reason to continue; adoption becomes a proxy for impact; delivery progress narrows the range of alternatives that still look practical.
The FDP evidence illustrates the problem without resolving the programme’s ultimate value. NHS England reports broad uptake—139 trusts live and 137 reporting benefits—and describes operational improvements. It also states that several before-and-after comparisons do not establish causality. The Office for Statistics Regulation has asked for clearer presentation, fuller communication of limitations and stronger statistical involvement. Adoption is therefore important evidence of reach, but it is not the same thing as proof that the observed benefits were caused by the platform. [S3] [S4]
This is precisely where a green delivery signal can become over-comforting. It says the commitment may be executable. It does not reconcile cost and benefit bases, price non-monetised public value, test causal attribution or compare continuation with credible alternatives.
The option point is the governance point
Value review matters most when a decision is still reversible. The FDP contract has an option point approaching in 2027, and the government has been reviewing the decision around continuation. That does not imply that exit is the right answer. It means that the board’s most consequential question is no longer only whether the programme can deliver. It is what evidence would justify the next tranche of commitment before switching costs and operational dependence rise further. [S7] [S8]
Boards often schedule value reviews according to reporting cycles. A stronger design schedules them around option points:
- contract breaks and extensions;
- material changes in cost, scope or benefit forecasts;
- architectural choices that increase vendor or platform dependency;
- transition from pilot to scaled operating model;
- new evidence that changes causal confidence;
- moments when a viable alternative will soon disappear.
At each point, the board should assess the remaining decision, not relitigate the entire history. Past expenditure is relevant only where it affects future options. The question is: from today, what additional commitment is required, what value can reasonably be attributed to it, what alternatives remain, and what becomes harder to reverse after approval?
A practical value-confidence test
A value-confidence judgement should be concise enough to govern, but rigorous enough to expose uncertainty. Six tests are sufficient.
Remaining commitment
State future cash cost, leadership attention, implementation capacity and opportunity cost. Separate these from sunk cost. Reconcile time periods and price bases before comparing costs with benefit forecasts.
Benefit causality
Distinguish activity, adoption, operational association and demonstrated causal effect. A platform can be heavily used and useful while the size of attributable benefit remains uncertain. Independent evaluation should define what evidence will change the decision, not merely add another report after the option has passed.
Strategic necessity
Identify value that may not be monetised: resilience, interoperability, clinical or citizen outcomes, strategic learning and the ability to coordinate a system. These benefits should be described and evidenced rather than treated as a balancing residual whenever the quantified case weakens.
Dependency and switching cost
Map which processes, data flows and capabilities are becoming dependent on the solution. Show how continuation, migration and exit costs change over time. Dependency is neither automatically bad nor automatically valuable; it is a governance fact that changes the option set.
Alternatives
Compare the current path with realistic choices: narrower scope, different sequencing, modular substitution, supplier renegotiation, internal capability or managed exit. An alternative that cannot be implemented within the available window is not a live alternative.
Reversibility
Name the next option date, decision owner and evidence deadline. A review that concludes after contractual or operational lock-in is assurance theatre.
The output need not pretend to mathematical precision. It should give a confidence level, the evidence basis, the dominant uncertainty, the strongest contrary interpretation and the action required before the next option closes.
The strongest case against a second rating
The sceptical argument deserves weight. A paired confidence model could duplicate existing business-case, benefits-realisation and assurance processes. It could create another coloured status that executives misunderstand. Cost forecasts may rise because adoption and scope expand; benefit estimates may become more conservative as evidence improves. Strategic infrastructure often produces system benefits that are difficult to monetise. Constantly reopening value could destabilise delivery and make long-term transformation impossible.
The FDP programme also has contrary evidence to a simple decline narrative. Its full business case passed formal approval. Uptake is broad. NHS organisations report operational benefits. An independent evaluation strategy is being developed, creating a route to stronger evidence. [S5] [S6]
These objections sharpen the model. Value confidence must not become a rolling referendum or a simplistic net-benefit traffic light. It should be triggered by material changes and real option points. It must treat non-monetised value explicitly, allow for the economics of scale-up and show whether uncertainty is reducing on a timetable that serves the decision. Most importantly, it must sit beside delivery confidence rather than compete with it.
The standard is not certainty. It is a decision good enough for the degree of irreversibility being accepted.
What a board should see
A board paper at an option point should fit the decision onto one joined view:
- Delivery confidence: current rating, scope, evidence date and principal execution risk.
- Value confidence: current judgement, comparable remaining cost, benefit evidence and material non-monetised value.
- Divergence: why the two judgements differ and whether the gap is widening or closing.
- Options: continue, rescope, pause or exit, with practical consequences and switching costs.
- Uncertainty: the fact most likely to change the recommendation and when better evidence will arrive.
- Decision: the commitment being authorised now, its owner and the next point of review.
For the FDP, such a view would preserve several truths simultaneously: delivery may be credible; adoption is material; official benefit claims require causal care; cost and benefit forecasts may not be directly comparable; strategic and clinical value may extend beyond monetised estimates; and the approaching contract choice deserves an explicit decision. None of those truths requires a premature verdict on the programme. Together, they require a better question.
Green is a beginning, not a conclusion
Programme governance is strongest when its signals retain disciplined meanings. Delivery confidence should tell leaders whether the current commitment can be executed. It should not be stretched into evidence that the commitment remains strategically and economically justified.
Value confidence supplies the missing judgement. Used at material changes and option points, it makes sunk cost, causal uncertainty, dependency, public value and alternatives visible before reversibility disappears. It also protects good programmes: a strong value case can justify perseverance through delivery difficulty, while a weak or uncertain one can prompt rescoping before more capability is locked into the wrong shape.
A green delivery rating is useful precisely because it is limited. The board’s responsibility begins where that limit becomes consequential.
Sources
- National Infrastructure and Service Transformation Authority — Major Projects Annual Report 2025–26 — 13 July 2026 — https://www.gov.uk/government/publications/nista-major-projects-annual-report-2025-26/nista-major-projects-annual-report-2025-26
- Financial Times — UK raises cost projections for Palantir’s NHS data platform — 10 August 2026 — https://www.ft.com/content/da888f06-b9bc-4d83-8020-e82f5d5fa609
- NHS England — FDP uptake and benefits — 12 June 2026 — https://www.england.nhs.uk/digitaltechnology/nhs-federated-data-platform/impact/fdp-uptake-and-benefits/
- Office for Statistics Regulation — NHS England Federated Data Platform: presentation and communication of information and performance metrics — 22 July 2026 — https://osr.statisticsauthority.gov.uk/news/nhs-england-federated-data-platform-fdp-presentation-and-communication-of-information-and-performance-metrics/
- Imperial College Health Partners — FDP Evaluation Strategy — August 2026 — https://www.imperial-consultants.co.uk/wp-content/uploads/2026/08/FDP-Evaluation-Strategy.pdf
- Department of Health and Social Care — Federated Data Platform programme Accounting Officer Assessment — 26 March 2026 — https://www.gov.uk/government/publications/department-of-health-and-social-care-accounting-officer-assessments/federated-data-platform-programme-accounting-officer-assessment
- UK Parliament — Federated Data Platform oral evidence briefing — 16 June 2026 — https://committees.parliament.uk/event/27486
- Reuters — UK reviewing Palantir’s NHS contract — 9 June 2026 — https://www.reuters.com/business/healthcare-pharmaceuticals/uk-reviewing-palantirs-nhs-contract-amid-pressure-use-break-clause-2026-06-09/