The 70% That Isn't: Auditing a Benefits Claim
A retailer's $30M supply chain program reports 70% benefits realization at month 18, and the program manager wants to present that figure to the board as evidence the investment is working. The number comes from vendor dashboards measuring "units processed per labor hour" at three pilot distribution centers. The baseline was captured during an industrial dispute, when throughput was abnormally low. Two of the five planned distribution centers have not deployed. Finance has not reviewed the methodology. The benefits owner, the VP of Operations, supports reporting 70%; the CFO, seeing soft cost savings, wants to suspend remaining funding.
1. How should the program manager test whether the 70% figure is defensible?
Read the worked answer
The program manager should treat this as a benefits audit, not a communications problem. Four tests:
- Baseline validity - a baseline captured during an industrial dispute is not a normal-operations baseline. If throughput was artificially depressed, every subsequent comparison is inflated. Reconstruct the baseline from a stable period or a comparable set of distribution centers.
- Attribution - how much of the gain is the program, versus market volume, seasonal mix, or the end of the dispute? Without a control group of non-deployed centers, the program manager should compare deployed against not-yet-deployed sites, which the two undeployed DCs conveniently provide.
- Coverage - three of five sites is a pilot sample, not program realization. Extrapolating pilot performance to full scope is a forecast, and should be labeled as one.
- Metric fit - units per labor hour is an operational productivity proxy. Is it causally linked to the benefits in the business case (cost per unit shipped, working capital, service levels)? If not, it's a leading indicator, not a benefit.
Have finance co-sign the methodology before any number reaches the board.
2. The CFO wants to suspend the program and the VP of Operations wants to report 70%. How should the program manager position the decision?
Read the worked answer
Both positions rest on the same unvalidated data, so the program manager's job is to separate the measurement question from the funding question.
- Give the board a restated figure with a stated confidence range, and explicitly label pilot-only results as pilot-only. A credible "22–35% realized, methodology under review" is worth more than an indefensible 70%.
- Point out that suspension is also a decision made on bad data. If the true realization is genuinely weak, suspension may be right - but the CFO should decide that on validated numbers, not on a dashboard artifact.
- Propose a decidable test: agree with finance what realization threshold at the next gate would justify continuing, and commit to measuring it with the corrected method.
- Keep roles clean: the VP of Operations is the benefits owner and is accountable for realizing and reporting benefits; the program manager is accountable for measuring them honestly. The program manager should not let the benefits owner's reporting preference drive the methodology.
Agreeing a threshold in advance is what converts a political fight into a governance decision.
3. What should change in the benefits realization plan going forward?
Read the worked answer
- Measurement method per benefit, signed off by finance, with formula, data source, frequency, and the accountable owner named.
- A defensible baseline, documented with the period and conditions it was drawn from, and a rule for re-baselining.
- Attribution approach - control or comparison sites, and an explicit statement of what is excluded.
- Leading vs. lagging indicators clearly separated, so operational proxies are never presented as realized benefits.
- A realization curve by site, reflecting that only two of five sites have deployed, with cumulative targets at each gate.
- A disbenefit and risk view - the industrial dispute and deployment delays belong in the benefits risk register.
The underlying principle: a benefits register that cannot survive an audit is a wish list. Benefits management is a measurement discipline with an owner, not a status slide.