Strategy

Making room for uncertainty

A project may look worthwhile until support costs rise or fewer people use it. Test the assumptions that could change your decision before committing.

A business leader choosing between a new service, better customer support and replacing an ageing system needs more than a promised return. Each proposal competes for money and the time of people who already have work to do. The useful question is what would make one proposal worth approving, and what would make the leader choose differently.

The sales team may expect more customers. Operations may expect fewer mistakes. The technology team may warn that the current system will soon lose supplier support. Put those claims alongside each other over the same period, showing the work and costs behind them. A single score can hide the very differences the approver needs to understand.

Compare spending now with the real alternatives

Start with the person making the decision, the deadline and the problem staff or customers face today. Compare the full proposal with a smaller change, waiting and keeping things as they are. Keeping the current arrangement still costs money if staff have to correct mistakes by hand or maintain old software. Waiting may also close a sales opportunity or leave the business using a system its supplier will no longer repair.

Count the work needed after purchase as well as the purchase itself. Someone must connect the new service to existing systems, train staff, check results and keep it running. If staff save an hour preparing an answer but spend an hour checking it, the business has not saved that hour. A forecast of extra sales also needs a reason to believe customers will buy and that the business can fulfil their orders.

Find the uncertainty that could change the decision

A team completing a recovery exercise is evidence of what happened under those test conditions. A supplier's promise is a commitment to check. A sales forecast is an estimate. Keep those distinctions visible. For an important estimate, show where it came from, the plausible lower and upper outcomes and why those limits were chosen. If nobody knows how much support a new service will need, explain how that gap affects the spending decision.

Ask what happens if customers take longer to adopt the service, support takes more staff time or an interruption lasts longer than planned. You do not need to attach a precise probability to every outcome to make this comparison useful. If a small change makes another option preferable, investigate that assumption before approval. More decimal places in the return estimate will not make the underlying uncertainty go away.

Use a trial to answer a spending question

A trial is useful when its result could change approval. In this example, the decision depends on whether staff get reliable answers and whether customer information stays restricted to the right people during real work. More users or faster answers would not settle those questions. Compare the trial's cost in money and staff time with the cost of approving the larger project without knowing the answer.

Also ask what the business can undo. It may be possible to switch off a service, but that will not retrieve customer information already copied elsewhere. Leaving may require moving records, retraining staff or paying a supplier. Describe those consequences before treating a trial as a harmless experiment. Legal duties and safety requirements still apply even when the estimated benefit looks attractive.

Write down what approval depends on

Keep a short record of the choice, the alternatives rejected and the assumptions that mattered. Say what evidence is needed before spending more, who can pause the work and when the decision will be reviewed. If a supplier changes its terms or the service needs much more support than expected, review sooner rather than waiting for the calendar date.

ILLUSTRATIVE · REVIEW STRUCTUREProposalOptions comparedReversing assumption recordedNamed owner + review triggerKeep rejected options and why they lost.What change would reverse the choice?Who acts, and what reopens the decision?ApproveDeferReject
Preserve the alternatives and the assumption that could change the choice. Whichever decision is reached, record an owner and a reason to reopen it. Illustrative example, not an engagement result.

Once the service is in use, compare what actually happened with what approval depended on. A good result can still follow a poorly supported decision, and a reasonable decision can still have a bad outcome. Separate what could have been known at approval from what changed later. Use that distinction to decide whether to continue, change the work or stop, and to improve the next spending decision.

Where this applies in our work: Investment evaluation.