Test the Product Business Case Where It Is Weakest

A product proposal shows a positive first-year return. The spreadsheet multiplies a target customer count by the subscription price, subtracts development costs and presents the difference as the reason to proceed. Before accepting the conclusion, the reviewer needs to know whether the customer count means paying accounts at year-end or the average number paying throughout the year.

That distinction alone can change the result. Sensitivity analysis makes uncertain assumptions visible and shows which changes would alter the decision. The useful output is a test or commitment the team can justify with the available evidence.

Consider a fictional maintenance-planning product for small facilities businesses. All figures in this article are invented. The model is deliberately limited so that its assumptions can be inspected; it is not a complete investment appraisal or a forecast for a real business.

Define the decision and the cost boundary

The proposed decision is whether to fund a first release for a specified customer segment. Compare it with continuing the existing service and with a smaller learning stage. If the options solve different problems, reconcile their objectives before treating their spreadsheet totals as comparable.

For the worked example, assume £36,000 of one-time implementation cost and £12,000 of fixed first-year operating cost. Assume each active paying account generates £30 a month and incurs £10 of variable service cost. Monthly contribution per account is therefore £20. Variable costs have already been deducted in that contribution figure; subtracting the same support cost again would understate the result.

These assumptions exclude items such as acquisition spending, tax and financing. Any real case would need to identify and include relevant incremental costs, the cash timeline and the organisation's appraisal method. The example tests a narrow calculation rather than declaring the investment attractive.

Use account-months when adoption changes over time

At an average of 300 active paying accounts across 12 months, the model has 3,600 account-months. Multiplying by £20 gives £72,000 of contribution. Against £48,000 of the specified implementation and fixed operating costs, the difference is £24,000.

A target of 300 accounts at the end of the year does not establish 3,600 account-months. If the average were 175 instead, the same calculation would produce 2,100 account-months, £42,000 of contribution and a £6,000 shortfall against the specified costs. Both scenarios could finish the year with 300 accounts; their earlier adoption paths would differ.

These are scenarios, not probabilities. The combined downside changes adoption, unit contribution and fixed cost together. Its result should not be described as a confidence interval or expected loss without a defensible probability model.

Find the assumption that changes the choice

In the initial simplified model, £48,000 divided by £20 gives 2,400 account-months. Spread evenly over a year, that is an average of 200 paying accounts. Below that volume, the specified costs exceed contribution; above it, they do not. This threshold ignores the excluded costs and the timing of cash, so it is not a complete break-even claim for the business.

The 2026 Green Book uses sensitivity analysis and switching values to examine how uncertain assumptions affect an option's value. Its framework concerns UK public spending and social value. The private-product calculation here borrows the question about changing assumptions; it does not claim to implement the Green Book's social-value appraisal or approval requirements. Green Book 2026, sensitivity analysis and switching values

A threshold helps the reviewer ask a sharper question: what evidence supports enough account-months at the assumed contribution? A list of interested organisations does not establish their purchase timing, continued use or willingness to pay. The team needs evidence about the inputs that sit closest to the decision threshold.

Choose evidence that can actually change the model

For adoption, investigate the buying process and the time between an expression of interest and paid use. For variable cost, observe the help required during a realistic trial. For retention, define what continued use means and inspect whether the product completes a recurring task.

Keep observations separate from the projection. If a trial involved substantial staff help, record that effort in the cost evidence. If participants were recruited from unusually enthusiastic customers, record that selection limit before generalising to the target segment.

The Government Service Manual recommends establishing a baseline, testing risky assumptions and revisiting estimated benefits as evidence develops. It is public-service guidance; these suggested research questions are an application to the fictional product. Measuring service benefits

Turn the uncertain case into a bounded decision

Check cash and capacity before claiming a benefit

The model's first-year difference does not show whether the business can fund implementation before subscription receipts arrive. Draw the cash timeline separately. A positive total at the end of a period can coexist with a funding gap earlier in it.

If an option releases staff time instead of generating subscription contribution, use the business-case guide's distinction between capacity and cash savings. Keep that valuation separate from the account-month sensitivity calculation here, and agree the treatment with finance.

A smaller trial can be the next decision even when the full proposal looks attractive. For this maintenance product, an observed buying cycle or measured support burden might change the case more than another refinement to the presentation. Save the model version and assumptions before the test, then show which inputs changed and what that means for the next commitment.