The audit standard an outcome must pass before it may be called delivered.
When we re-audited 312 outcomes that programmes had reported as delivered, 44% failed at least one basic test: the baseline had quietly moved, the benefit was counted twice, or the result did not survive twelve months. Transformation statistics rest on declarations. That holds for the industry’s published figures and for the numbers most organisations report to their own boards.
The standard replaces declaration with verification. An outcome may be called delivered only after it passes three tests against evidence an independent reader can audit. The tests are deliberately few and deliberately hard to argue with. Every figure we publish has passed them. That is what our numbers mean, and it is also the limit of what they mean: verification shows that a benefit happened, that it belongs to this programme, and that it lasted. It says nothing about whether the benefit was worth pursuing. That is a different argument.
The month-0 baseline was frozen before work began and is under change control. The outcome is measured against that baseline and never against a later re-forecast.
Test 1The benefit has exactly one attribution line. No other initiative, budget or business-as-usual movement claims the same value.
Test 2The outcome still holds twelve months after delivery was claimed. Results that decay were project artefacts, not delivered value.
Test 3Verification is performed by someone outside the delivering team, the evidence trail is written so that a later reader can re-audit it, and an outcome that fails any test returns to the programme as open work. The client supplies the baseline documentation and the finance extract. There is one case we decline outright: where no baseline was frozen before the work started, there is nothing defensible to measure against, and we will not reconstruct one after the fact so that a result can be signed off.
Transformation programmes, as the exit gate of every Release phase in Draw · Hold · Release.
Annual cohort research, where no outcome enters the Markham dataset until it has passed the standard.
Board and investor reporting, where delivered value needs to survive diligence rather than decorate a slide.
We have applied the standard to more than three hundred outcomes since 2023, and what it mostly produces is a gap between what was reported at the time and what actually held:
The Institute stewards the standard and revises it annually rather than by application count, since a standard people cite in board papers should be stable enough to cite years later. Tests, templates and pass criteria are published in full and may be applied without licence. The survival test used to be advisory. Making it mandatory is the one revision that made the standard harder to pass, and it stands because a result that decays was never delivered value.
Cite as: Markham Institute, “The Verification Standard”, MKM-F-003, v2.0 (2025).