The Benefits That Vanished
Eighteen months after go-live of a new planning platform, the business case's US$14M annual benefit has been validated by finance at US$4.2M. Regional inventory is down, but expedite freight is up enough to consume much of the gain, and nobody owns the net number. KPI definitions have changed three times, so the original baseline no longer reconciles with current reporting. The risk register was last reviewed seven months ago; since then a Tier-1 supplier has become single-source on a critical component and two of the three solution architects have left. The continuous-improvement backlog holds 210 items with no scoring, and the steering committee has started asking whether to fund phase two.
1. How should the team reconstruct benefit realization so the gap becomes a decision input rather than a reporting argument?
Read the worked answer
The gap is not a reporting defect; it is information about whether the transformation worked. Treat it that way.
- Freeze a baseline now and lock the definitions. Document what each KPI meant at each of the three definition changes and restate history onto one definition. Publish it as the reference point for everything that follows. Continued redefinition is how benefits disappear.
- Attribute with a counterfactual. For each benefit line, state what would have happened without the platform (volume, price, mix, demand volatility). Inventory down in a soft demand year is partly a market effect, and finance will discount anything that ignores this.
- Net the offsets. Expedite freight up is a real cost of the new planning behaviour, likely from tighter inventory targets pushing risk into transport. Report net benefit by P&L line, not gross savings per function. A benefit that is offset elsewhere is not a benefit.
- Assign an owner to every line, including the net number. Unowned benefit lines do not materialize by themselves.
- Classify hard versus soft, and state the realization confidence for each.
- Restate the case and present three options to the committee: recover a defined subset with named actions, re-scope phase two to the recoverable lines, or stop and harvest.
The question to answer explicitly: is the shortfall a delivery problem, a business-case problem, or a design problem? Each has a different response.
2. What should be done about the stale risk register, and how should supply concentration and key-person loss be governed from here?
Read the worked answer
A register reviewed seven months ago, in a programme that has since changed its supplier base and lost two architects, is not a risk register - it is a document.
Refresh approach:
- Re-baseline by interview, not by workshop, with the line owners of each process. Risks are owned where the work happens: the single-source component belongs to the category manager and the operations lead, not to the PMO.
- Score on likelihood and impact in money and service, with a named owner, a mitigation action, a trigger and a review date. Anything without an owner and a date is deleted.
The two live risks:
- Single-source critical component. Qualify a second source with a dated plan even where dual-sourcing costs more; quantify the exposure as days of line stoppage and revenue at risk. Add a contractual and inventory buffer decision. Report it monthly until the second source is qualified.
- Key-person loss. Two of three architects gone means undocumented design decisions. Commission knowledge capture on the highest-consequence configurations, and add an explicit succession line for every critical role, with vendor support terms reviewed.
Cadence: monthly operational review, quarterly steering review with the top five risks, and a standing rule that a risk whose trigger fires escalates automatically without waiting for the cycle.
3. How should the 210-item continuous-improvement backlog be handled so phase two can be decided on merit?
Read the worked answer
An unscored backlog of 210 items is a way of avoiding decisions, and it will be used either to justify phase two (see how much demand) or to bury it (see how little progress). Neither is evidence.
Triage in one pass:
- Score each item on value and effort, where value must map to a line already in the benefits register. Items that improve a real KPI stay; items that only improve tidiness are parked. Expect the list to compress sharply.
- Separate defect from enhancement. Post-go-live defects are warranty obligations under the contract and should be counted against the programme, not presented as improvement demand.
- Cap work in progress. Fund a small number of items fully rather than many partially; parallel partial work is where improvement programmes lose benefit.
- Publish the top ten with owners and dates, and the reason for parking the rest.
Then frame the phase-two decision on three questions: what is the recoverable benefit from completing and stabilizing what exists; what is the incremental benefit of new scope, with confidence ranges; and what is the capability risk if the architects are replaced mid-flight. Present recovery and expansion as separate decisions with separate cases, so the committee cannot approve expansion as a way of avoiding the recovery conversation.