At a glance
- Situation
- Large enterprise, 18 months into a multi-year transformation. $84M authorized, $46M spent, $38M remaining, 61 active workstreams.
- Complication
- $34M of claimed annual benefit against $9M modeled as Finance-validatable — an initial validatable rate of 26.5% — while delivery reporting stayed green or amber.
- The decision
- Accelerate, continue, rescope or stop — with $38M of authorized spend still ahead and no evidence base to decide on.
- Entry point
- Transformation Oversight. SORT, DO and PROVE were the most visible stages.
- Result
- $18M of remaining authorized investment protected or redirected — capital, not savings; modeled Finance-validatable annualized benefit from $9M to $28M — 82.4% of the original headline case.
It did not look like a disaster
This is the difficult category of intervention, because nothing in the program was obviously failing. Most workstreams had plans. Governance meetings happened on schedule and were attended. Milestones were being completed at a respectable rate. Status reporting was largely green, with amber where it should have been.
Judged as delivery, the program was performing. A steering committee reviewing it would have found little to escalate, and eighteen months of that reporting had built genuine confidence.
Judged as transformation, it was not performing at all. $9M of $34M in claimed annual benefit could be tied to established financial outcomes — a validatable rate of 26.5% — and the remaining $25M consisted of claims that had been reported but never reconciled to anything.
Delivery status answered a different question
The program was measuring and reporting the things programs measure: schedule adherence, milestone completion, budget consumption, risk and issue counts, and workstream RAG. Each is a legitimate delivery measure. Collectively they answer whether the work is progressing.
The question the executive committee now faced was different: with $38M of authorized spend remaining, should it be spent, and on what. Delivery status cannot answer that, because a workstream can be perfectly on schedule toward an outcome that will not arrive.
Do we accelerate, continue, rescope, or stop — and on what evidence?
Without a value-based read, the default answer to that question is always continue. Work proceeds because it has started, capacity flows to whatever is already in flight, and the decision is made by inertia rather than by anyone.
Why the usual options are both wrong
Two responses are typically offered at this point. The first is to press on and accelerate, on the grounds that the investment is largely committed and the benefits are back-loaded. The second is to stop the program and restart the planning from scratch.
Acceleration compounds the problem when the issue is that value cannot be traced. Moving faster along an unevidenced path produces more spend and no more proof, and it uses up the remaining authorization that would otherwise fund the correction.
A restart is equally wrong, and it is the misreading of zero-based thinking that this case is useful for correcting. Zero-based means refusing to inherit assumptions no one can defend. It does not mean discarding eighteen months of work that can be defended — and in this program a substantial portion could be.
Entering in flight is not a reason to restart from zero
The read tested the program against the outcomes it was funded to produce, workstream by workstream, and then re-decided the remaining investment on that basis.
Which required outcomes still matter
Eighteen months is long enough for the business context to move. Some outcomes in the original case were still the right ones; others had been overtaken, and the work pointed at them was proceeding on an obligation that had lapsed.
Which work remains necessary
For the outcomes that still mattered, the question was which of the 61 workstreams were genuinely required to produce them — as distinct from workstreams that had been scoped in at the outset and never re-examined.
Which initiatives still have a credible path to value
Each workstream was assessed for whether a route to evidenced value still existed from where it now stood. Sunk spend was explicitly excluded from that assessment; the only relevant question was what the remaining investment would produce.
Where dependencies and requirements were missed
Several workstreams were progressing well against a plan that assumed a dependency nobody owned. Those gaps were named and either funded or the dependent scope was stopped.
What value can be defended
Finally, the claimed benefit case was reconciled: what could be evidenced now, what could be evidenced later with the remaining investment, and what should never have been in the case.
What was stopped, rescoped and rebaselined
| Line | Value | Basis |
|---|---|---|
| Claimed annual benefits | $34M | At intervention |
| Modeled Finance-validatable at intervention | $9M | 26.5% validatable rate |
| Workstreams stopped | 11 | Of 61 |
| Materially rescoped | 14 | |
| Rebaselined | 19 | |
| Continued substantially intact | 17 | |
| Investment released by stopped work | $11M | Of the $38M remaining |
| Reduced or rescoped future spend | $7M | |
| Remaining investment protected or redirected | $18M | Of $38M authorized and unspent |
| Modeled Finance-validatable annualized benefit, twelve months after reset | $28M | 82.4% of the original headline case |
How the $9M became $28M
| Twelve-month value bridge | Annualized | Basis |
|---|---|---|
| Modeled Finance-validatable at intervention | $9M | Already reconciled to financial outcomes |
| Recovered from rescoped and rebaselined work | +$9M | Claims that existed but could not previously be evidenced |
| Newly proven from work continued intact | +$12M | Benefit arriving under evidence requirements set at the reset |
| Claims retired or reduced | −$2M | Removed from the case rather than carried unproven |
| Modeled Finance-validatable annualized benefit | $28M | Twelve months after reset |
Attribution: the intervention did not create $19M of benefit. Most of the movement is benefit that already existed in the program and could not be evidenced — the reset established the evidence path, retired what could not be proven, and redirected remaining investment toward the outcomes that still mattered. The portion attributable to the intervention itself is the $9M recovered from rescoped work and the $2M correctly retired; the $12M newly proven would substantially have arrived from work already under way.
The $18M is not described as savings. It is remaining investment protected or redirected — a decision about where unspent authorization should go.
Smaller in places. Stronger in others.
Eleven workstreams were stopped, fourteen materially rescoped and nineteen rebaselined; seventeen continued substantially intact. Roughly a quarter of the program was retained unchanged, which matters: a reset that stops everything is not a reset, it is a cancellation.
The $18M released by stopping and rescoping was not returned as a saving. It was redirected within the authorization to the outcomes that still mattered, including the dependencies that had been missed in the original plan.
Reporting changed at the same time. The program continued to report delivery status, but the primary governance view became Finance-validatable value against the committed case — which is the view the executive committee had been missing for eighteen months.
The results
Twelve months after the reset, modeled Finance-validatable annualized benefit reaches $28M, against $9M at intervention and $34M originally claimed — 82.4% of the original headline benefit case, from a program that had been validating at 26.5%. The bridge above shows where that movement came from, and how much of it is attributable to the intervention rather than to work already in flight.
Modeled Finance-validatable annualized benefit twelve months after the reset — 82.4% of the original headline case. Finance validation is the proof standard the benefit is held to, not a historical event this modeled case reports.
The story is not that a failed program was rescued. Much of the work was sound, and the delivery organization had been executing competently against the plan it was given. What changed was the standard by which continuing investment was decided.
Why the result held
The program was re-decided against value rather than schedule, so the reporting that followed answered the question the board was actually asking. Once that view existed, later decisions were straightforward.
The retained workstreams also carried evidence requirements that had not been specified originally. That is why the validatable value rises rather than merely being restated: the work that continued was now able to prove what it produced.
What this means if you are 18 months in
Green status and unvalidated benefit can coexist indefinitely. If nobody has reconciled the benefit case, the RAG report will keep looking reasonable.
Exclude sunk spend from the decision. The only question that matters is what the remaining authorization will produce.
A reset is not a restart. Expect a meaningful share of the program to survive unchanged — and be suspicious of a read that stops everything.
If more than half of a claimed benefit case cannot be tied to an established financial outcome, the program is not being governed. It is being reported.
Evidence basis
Execution performance and transformation value are separate measurements, and the research treats them that way.
- PMI, Pulse of the Profession 2024: The Future of Project Work. Reports an average project-performance rate of 73.8% across all respondents, measured as the mean share of completed projects that met business goals — execution can be reasonably strong while broader transformation-value questions remain open.
- PMI, Pulse of the Profession 2026: Driving Success in Complex Projects. Reports that project professionals who manage complexity effectively increase the likelihood of project success by five times, with success defined around value relative to effort and expense rather than schedule alone.
- McKinsey, Losing from day one (December 2021 global survey, n=1,034). Reports that 55% of transformation value loss occurs during and after implementation. Cited as context.
ETEGY · ZBT in Practice · Case 06 · In-flight recovery · etegy.com