What is Sensitivity Analysis?
Short answer: Sensitivity analysis measures how a model result changes when selected assumptions change while the calculation structure and other inputs remain fixed or deliberately paired. It does not predict which outcome will occur and does not assign probability. Its purpose is to show which assumptions drive a valuation, financing or investment decision and how much room exists before the decision changes.
A one-way sensitivity changes one input, such as revenue growth. A two-way table changes two inputs, such as discount rate and terminal growth. Scenario analysis changes a coherent set of variables together, while probabilistic simulation assigns distributions and relationships. These approaches should not be confused. Sensitivity ranges need evidence from historical volatility, contracts, market data or operating plans. Changing highly correlated variables independently can create impossible combinations. A table with many cells can also hide which cases are actually relevant to the board's choice.
How it works
Begin with a validated base model and identify the output that informs the decision. Rank uncertain inputs by impact and controllability. Choose plausible ranges and increments, preserving units and formula links. Recalculate the entire model for each case rather than overwriting the final output. For linked assumptions, use named scenarios with internally consistent revenue, margin, working capital and financing. Record the break-even value at which a covenant, minimum return or purchase-price limit is reached. Reconcile the base-case cell to the approved model and review non-linear effects or circularity.
Sensitivity change = recalculated output under varied assumption - base-case output
Example
A valuation varies maintainable EBITDA and the selected EV/EBITDA multiple. EBITDA cases are 1,700, 2,000 and 2,300, while multiples are 6.0x, 7.0x and 8.0x. At 6.0x, enterprise values are 10,200, 12,000 and 13,800. At 7.0x they are 11,900, 14,000 and 16,100. At 8.0x they are 13,600, 16,000 and 18,400. With net debt of 3,000, each equity value is 3,000 lower. A proposed equity price of 12,500 therefore requires enterprise value of 15,500, equivalent to 7.75x base EBITDA or about 6.74x upside EBITDA. The table makes that dependency directly recomputable.
Why it matters
Boards use sensitivity analysis to see whether a transaction, financing or plan is robust or relies on a narrow assumption. Investors use it to set return thresholds and identify diligence priorities. Lenders use it to test covenant and liquidity headroom. In negotiation, a break-even table can convert disagreement about one forecast into a clear price or structure discussion, such as an earn-out. The analysis is most useful when it leads to action: additional diligence, a lower price, more equity, a covenant buffer or a contingency plan.
A sensitivity is only as reliable as the underlying model and selected ranges. It does not capture omitted risks, probabilities or management reactions unless those are explicitly modelled. Holding other inputs constant can be unrealistic, while changing everything at once can obscure causation. Discount rates, terminal growth and multiples must remain economically coherent. Avoid including mathematically invalid cells or implying equal likelihood through symmetric formatting. Preserve formulas and audit trails, disclose the valuation date and basis, and use professional judgement when assumptions have accounting, tax, legal or market dependencies.