Draft for Tamir's review. Not published.
Find where the benefit stops
Lay each assumption in the case against what happened: output, uptime, labor, ramp-up time. Mark each assumption's source as a supplier claim, your own data or an estimate. Then follow the path from system to saving, such as meter readings that arrive, are valid and reach billing, as separate numbers.
Better metrics are not always cash. Reconcile reported utilization with costs the business can actually change under current arrangements. Count the manual work that still runs beside the system at peak, and check whether planned staffing cuts rest on verified service results.
Assess accountability fairly
Separate what was knowable at approval, what was decided during the program and what was outside anyone's control. Accountability attaches to the decisions. Name them, who made them and on what information, and include the board's own decisions.
Measure what the program cost beyond its budget, such as customers lost during disruptions, and present it with the budget figure. Write the lessons as requirements for the next program.
Depending on your seat
If you're on the board, require benefits reporting at the same cadence as delivery: each benefit with its business owner, measure, current value and date. Benefits with no owner outside the program are the ones that never arrive. Set a post-investment review at a fixed date for every capital project.
If you're the CEO, understand the failure before you fund another project or hire the next leader. Secure the documentation, contracts and vendor knowledge before people leave. Then fund only the gap you found.
What to check before you decide
- Lay the business case assumptions against the actual results, line by line.
- Mark each assumption's source: supplier claim, your own data or an estimate.
- Measure each step from system to saving separately, and find where the benefit stops.
- Separate freed capacity from costs the business can actually change.
- Name a business owner, measure and date for each benefit, and remove benefits nobody owns.
- Reconstruct the major decisions with dates, makers and the information they had.
- Set a post-investment review at a fixed date for the next program.
Questions people ask
We wrote off a failed technology project, how does the board assess whether its own governance was adequate?
Reconstruct what the board saw and decided at each point, against what the project's state actually was, and find where the board could have known and did not ask. The lessons are usually about reporting content, gate reviews and independent evidence, which the board can change. It depends on what the board received and on whether anyone outside the project ever reviewed it.
Our automation investment delivered a fraction of its business case, how does the board assess the case and improve the next one?
Compare the case's assumptions with what happened, line by line: output, uptime, labor, ramp-up time, and find which assumptions came from the supplier and which from the company's own measurements. Then require the next case to show its sources and a ramp-up based on evidence. It depends on whether the shortfall is in the equipment's performance or in the assumptions around it.
Our core migration finished late and over budget, how does the board assess management's accountability fairly?
Separate what was knowable at approval, what was decided during the program and what was outside anyone's control; accountability attaches to the second group. A fair assessment names the decisions, who made them and on what information, and sets requirements for the next program. It depends on the program's record and on what the board itself approved along the way.
Our program reports delivery progress but never benefits, what should the board require?
Require a benefits report alongside delivery: each benefit in the case with its owner in the business, the measure, the current value and the expected date. Benefits that have no owner outside the program are the ones that never arrive. It depends on whether the business case's benefits were defined with measures and on who owns them now.
Our IT project failed and the CIO resigned, what should I do with the project before hiring the next CIO?
Understand the failure before hiring, because the next CIO's job description depends on it: a vendor problem needs a negotiator, a scope problem needs a product mind, a team problem needs a builder. Keep the project in a defined hold with the vendor obligations clear. It depends on what the project was supposed to deliver and on whether any part of it can be used.
Smart meter rollout is done but the promised savings haven't appeared because billing can't use the data, what happened and what now?
Rollouts usually succeed on installation and fail on the data path from meter to billing, which nobody owned. Find the exact gap: data arriving, data quality, billing system readiness or process change. Then fund only the gap. It depends on where the data stops today and on whether the billing system was ever in the rollout's scope.
Our core migration is done but customers left during the disruptions, how do I account for it and learn from it?
Measure the churn and service complaints against the migration's timeline, and separate what the migration caused from what would have happened anyway. Then write what the next program will do differently about customer impact. It depends on whether customer impact was ever a measure in the program and on what the data shows.
Should our clinic reduce call centre staffing after buying voice AI?
Base staffing decisions on verified service outcomes and remaining work. The business case depends on call complexity, escalation demand, accessibility, and responsibilities approved by clinical and operational owners.
Why does new scheduling software leave our transport staffing costs unchanged?
Reconcile operational metrics with costs the business can actually change. Further investment depends on working arrangements, dispatch constraints, and whether freed capacity has a productive use.
Should we keep warehouse automation that requires manual work every peak season?
Evaluate the combined automated and manual operating model. Renewal or expansion depends on peak constraints, fallback effort, and whether a narrower automated scope produces credible value.
How I can help with this decision
- Ask or talk (Free)
- I give my view on where cases like yours usually break, and which numbers show your gap.
- Review (Pay if it was worth it)
- I write an independent post-investment review of the case against the results, the benefit path and the decisions behind it. I recommend what closes the gap, who owns it, and what to require of the next case.
- Retain (When it makes sense)
- I stay available to review the benefit reports, and the next cases against the new requirements.