Sensitivity analysis: identify the assumption that changes the decision.
Answer:
Sensitivity analysis is not varying every parameter by ±10%. It must find plausible assumptions that change the sign, ranking or feasibility of a decision. Parameters are varied according to their uncertainty and, when needed, jointly.
Method and decision:
Publish the base scenario, outcome, constraints and alternative. Separate estimated parameters, controlled decisions, constraints and external events because their ranges and treatment differ. Use a tornado to locate sensitive parameters, then linked scenarios to test plausible combinations. Compare test cost with avoidable loss if the central assumption is false. Adopt a recommendation only when it remains acceptable in plausible scenarios, or fund information that reduces uncertainty in the decisive parameter.
Worked example:
Noroa reallocation gives +€122,000 centrally. Plausible media-elasticity variation moves the result from −€92,000 to +€71,000; margin moves it from −€64,000 to +€58,000. Under prudent elasticity alone, gain remains +€30,000. Combining prudent elasticity with low margin makes the result −€34,000. A six-week test costs €18,000 and can halve elasticity uncertainty. Do not deploy the target directly; fund the pilot because its cost is below the potential loss in the adverse joint scenario.
Limits:
Local analysis may miss non-linearities or strong interactions. Scenarios cannot cover unimagined events. Ranges that are too narrow turn sensitivity into confirmation of the central scenario.