A base case is one possible future. If a recommendation changes when one uncertain input moves slightly, the decision needs that fact on the first page. Sensitivity analysis makes uncertainty visible by changing selected assumptions while holding the rest of the model constant.
Start with a small model
Imagine a fictional program with $1,200 of new annual fixed cost. Each distribution moved away from the old method avoids $1.40 of variable cost but adds $0.40 of recurring administration. Net saving per moved distribution is therefore $1.00. The annual break-even volume is 1,200 distributions: fixed cost divided by net unit saving.
At 800 moved distributions, the model shows an annual net cost of $400. At 1,200, it breaks even. At 1,600, it shows a $400 saving. These are arithmetic scenarios, not Corpay pricing, adoption forecasts or a claim that employees should choose a particular method.
Change the unit saving too
Volume alone is not enough. If net unit saving is only $0.50, break-even rises to 2,400 moved distributions. If it is $1.50, break-even falls to 800. A proposal that looks robust under the central $1.00 assumption may fail when retained labor or an overlooked minimum charge reduces the avoidable cost.
- At 800 moved payments: net annual result is minus $800 at $0.50, minus $400 at $1.00, and zero at $1.50 per payment.
- At 1,200 moved payments: net annual result is minus $600, zero, and plus $600 respectively.
- At 1,600 moved payments: net annual result is minus $400, plus $400, and plus $1,200 respectively.
Do not make impossible scenarios
Check the scenarios against the available volume, voluntary participation and retained payment methods. A break-even point of 2,400 distributions is not informative unless that many distributions could plausibly move. If annual eligible volume is only 1,800, the low-saving case cannot reach break-even within the stated boundary. Say so instead of extending the chart beyond operational reality.
Also keep correlated assumptions together when appropriate. Higher participation may require more communication work, rather than leaving administration perfectly flat. A shorter rollout may reduce first-year volume and increase transition intensity. A one-variable test is a diagnostic, not a complete forecast.
Turn sensitivity into a research plan
Which unknown is most likely to change the decision? If the case is highly sensitive to handling time, observe that work before negotiating cosmetic details. If a contract minimum dominates, request a written explanation. If participation drives the result, avoid assuming an outcome that workers have not chosen.
A clear recommendation can be conditional: proceed only if documented unit savings and a realistic voluntary payment mix exceed a stated threshold. Cost is still only one consideration. Access, payment choice, legal duties, reliability and worker experience should not be converted into arbitrary dollar values merely to force them into this spreadsheet. Review those dimensions alongside the model.
Read the threshold as a question
The break-even result of 1,200 moved distributions can be translated into a practical research question: can the organization realistically and lawfully reach that annual volume with the assumed net unit saving? If there are 100 eligible recipients and 26 runs, the theoretical maximum is 2,600 distributions, but voluntary choices, launch timing and turnover reduce the relevant opportunity. A theoretical maximum is not a forecast and should not be used as the central case.
For illustration, 1,200 divided by 26 is about 46.2 recipients paid through the changed method in every run. Because people and payments are discrete, that average is only a planning translation. Forty-seven full-year recipients would produce 1,222 distributions, while a midyear switch requires a different calculation. This interpretation helps a reviewer see whether the threshold is close to or far from plausible operating conditions without pressuring anyone to choose a method.
Test a downside combination
A one-variable sensitivity can understate risk when uncertainties move together. Suppose net unit saving falls to $0.75 and annual fixed cost rises to $1,500 in a fictional downside case. Break-even becomes 2,000 distributions. At 1,600 distributions the result is minus $300, rather than the plus $400 central result. The comparison shows why two individually modest changes can reverse the decision when combined.
Do not assign probabilities to these scenarios without a basis. A low, central and high case is not automatically a statistical distribution. If the team does have a defensible range from measurements or contract options, document it. Otherwise present scenarios as stress tests: what the result would be if these conditions occurred. That language preserves their usefulness without pretending to know how likely each future is.
Finally, test the sign of the denominator. If the proposed method costs more per distribution than the avoided variable cost, net unit saving is zero or negative. Dividing fixed cost by that value does not produce a meaningful positive break-even volume. More volume would not rescue the model under those assumptions. State that result directly and review whether nonfinancial reasons justify the change separately.