Excel Solver Sensitivity Report: How to Read It

How-to basics · Updated November 21, 2026

An Excel Solver sensitivity report shows how changes to a linear optimization model’s objective coefficients and constraint limits could affect its solution. To create one in Excel for Microsoft 365 on Windows, solve a continuous linear model with the Simplex LP method, then select Sensitivity in the Solver Results dialog.

What the sensitivity report tells you

Solver finds values for decision variables—such as how many units to make or how to allocate a budget—that maximize or minimize an objective while meeting constraints. The sensitivity report adds context to that answer: it estimates which limits are valuable, how much an objective coefficient can change before the current solution changes, and how sensitive the result is to adjustments in constraint right-hand sides.

It is not a forecast, a probability analysis, or a guarantee that a revised model will have the same answer. Its ranges describe changes around the current optimum under specific linear-programming assumptions. If you change inputs outside those ranges, or change several inputs together, rerun Solver rather than relying on the existing report.

Prepare a model Solver can analyze

A typical model has a cell for the objective, cells for decision variables, and formulas that calculate resource use or other results. For example, a production plan might use two changing cells for quantities of products A and B. A profit formula calculates total profit, while separate formulas calculate labor hours and material use. Constraints limit those resource totals, and the decision cells are usually restricted to nonnegative values.

In Windows Excel, Solver is an add-in. If the Solver command is missing, enable it through File > Options > Add-ins. At the bottom, choose Excel Add-ins next to Manage, select Go, and check Solver Add-in. Then open Data > Solver. In the Solver Parameters dialog, set the objective cell, choose Max or Min (or a target value), identify the changing variable cells, and enter the constraints.

For a sensitivity report, choose Simplex LP as the solving method. The model must be linear: the objective and constraint formulas need to change linearly with the decision variables. Products of changing variables, variable exponents, and many nonlinear formulas do not meet that condition. Also check that each constraint reflects the intended business rule; sensitivity results cannot compensate for an incorrectly specified model.

Generate the report in Excel

  1. Open Data > Solver and confirm the objective, variable cells, and constraints.
  2. Choose Simplex LP from the solving method list, then select Solve.
  3. When Solver displays Solver Results, keep the solution if appropriate and select Sensitivity from the Reports list.
  4. Select OK. Excel creates a new worksheet containing the report.

Solver’s available reports depend on the model and method. The Sensitivity report is generally offered for a continuous linear model solved with Simplex LP; it is not the right report for an integer or binary model, and nonlinear models require different interpretation. If Sensitivity is absent, check the solving method and whether the model contains integer constraints or nonlinear formulas. Solver for Excel on Mac has a different add-in setup and may present interface details differently; the report’s concepts are the same when the appropriate Solver feature is available. Excel for the web does not provide the same built-in Solver add-in workflow as desktop Excel.

Read the Variable Cells section

The Variable Cells section reports one row per changing cell. Final Value is the value in the solution. Objective Coefficient is that variable’s contribution rate in the objective formula—for example, profit per unit. Reduced Cost and the allowable increase and decrease describe how the objective coefficient can change before the current optimal basis changes.

For a variable with a positive final value in a standard linear program, reduced cost is normally zero. A variable at zero may have a nonzero reduced cost: broadly, its objective coefficient would need to improve enough before that variable becomes attractive in the solution. The sign and interpretation depend on whether the objective is a maximization or minimization and on Solver’s convention, so use the displayed value with the model’s direction in mind. Do not read reduced cost as the variable’s actual cost or as a penalty charged by Solver.

Allowable Increase and Allowable Decrease give the interval over which an individual objective coefficient may change while the current optimal basis remains optimal, with other model inputs held fixed. Within that interval, the decision-variable mix generally remains optimal, although the objective value changes as the coefficient changes. At a boundary or when multiple optima exist, details can be less intuitive. Outside the range, solve the model again.

Interpret constraints, slack, and shadow prices

The Constraints section includes Final Value, Shadow Price, Constraint Right-Hand Side, and allowable increase and decrease. Final Value is the calculated left-hand side of the constraint at the solution; the right-hand-side column is the limit entered in the model. For a binding constraint, the final value meets the limit. A nonbinding constraint has slack (for a less-than-or-equal-to limit) or surplus (for a greater-than-or-equal-to limit).

A shadow price estimates how much the optimal objective value would change for a one-unit increase in a constraint’s right-hand side, while the other inputs stay fixed and the change remains in the report’s allowable range. Its units are objective units per constraint unit: for example, dollars of objective per additional labor hour. The sign depends on the objective direction and the constraint type. A nonbinding constraint in a conventional linear program often has a shadow price of zero, because a small increase in its limit does not improve the current objective.

A shadow price is not automatically the amount you should pay for more capacity. It describes a marginal change in the mathematical model, within a stated range; it does not include implementation costs, uncertainty, or practical constraints omitted from the model. Compare the modeled value with real-world costs and conditions before acting.

Use allowable ranges correctly

For a constraint, the allowable increase and decrease indicate how far its right-hand side can move, one at a time, while the current basis and shadow-price interpretation remain valid. If a binding labor limit has a shadow price of $2.50 per hour and an allowable increase of 20 hours, the report suggests that increasing the limit by 10 hours would improve the objective by about $25, assuming all else stays fixed and the model remains valid. At an increase of 20 hours, the estimate reaches the stated boundary; a larger change calls for a new solve.

The phrase “one at a time” matters. Sensitivity ranges usually do not establish what happens when several coefficients or limits change simultaneously. Even if each individual change falls within its own reported range, their combined effect may differ. For multiple changes, use the report as a guide to identify important inputs, edit the model, and run Solver again.

A small illustrative example

Imagine a factory choosing quantities of two products to maximize contribution while subject to labor and material limits. Suppose a hypothetical Solver report shows that the labor constraint is binding, its shadow price is $2.50 per labor hour, and its allowable increase is 20 hours. If management considers adding 8 hours, the linear sensitivity estimate is an objective improvement of $20: 8 multiplied by $2.50. That is an estimate from the model, not a promise of extra cash profit; it assumes the same model and all other inputs remain unchanged.

Now suppose the report shows 12 unused hours on the material constraint and a zero shadow price. Adding more material capacity would not improve the current solution by itself, because material is not the limiting resource at that point. The production mix may be constrained by labor instead. These two lines help distinguish a scarce modeled resource from one that currently has room to spare, but they do not tell you whether the factory can actually obtain labor or material at a particular price.

Common problems and misconceptions

  • No Sensitivity option: Check that Simplex LP is selected and that the model is continuous and linear. Integer or binary restrictions can make this report unavailable.
  • Nonlinear formulas: A formula that multiplies two decision-variable cells is nonlinear. A report based on a linear model is not suitable for interpreting such a relationship.
  • Confusing shadow price with total value: It is a marginal estimate per unit, not the total value of the resource or a recommended purchase price.
  • Treating allowable changes as simultaneous: The reported ranges usually vary one input at a time. Multiple edits require a fresh Solver run.
  • Ignoring model units: A shadow price per hour is not comparable directly with a price per kilogram. Check the constraint’s units and the objective’s units before interpreting it.

Another common misconception is that the report tells you how certain the solution is. It does not provide confidence intervals or probabilities. Its conclusions are conditional on the model’s structure and inputs. If demand, prices, yields, or capacity are uncertain, scenario analysis or a separate uncertainty model may be more useful.

When to use another analysis

Use the sensitivity report when you have a linear optimization model and want to understand local trade-offs around its optimum. For a model with yes-or-no decisions, whole-number quantities, or nonlinear relationships, inspect Solver’s solution and use suitable scenarios instead of expecting the continuous LP sensitivity interpretation to apply. You can change one assumption at a time, record the resulting objective and decision values, and compare cases. For larger or more specialized optimization problems, dedicated optimization software may offer tools beyond Excel Solver.

Keep a copy of the original assumptions and solution when exploring changes. A report worksheet is tied to the model state that produced it; after changing constraints, coefficients, formulas, or decision cells, solve again and generate an updated report. This makes it easier to distinguish an old sensitivity result from the results for a revised model.

Frequently Asked Questions

What does a sensitivity report show in Excel Solver?
It summarizes how an optimal linear solution relates to objective coefficients and constraint limits. It includes reduced costs, shadow prices, and allowable changes for the model’s variable cells and constraints.

Why is the Sensitivity report unavailable in Solver?
The usual cause is that the model is not a continuous linear model solved with Simplex LP. Check the solving method, remove or reconsider integer and binary restrictions if appropriate, and confirm that the formulas are linear.

What does a zero shadow price mean?
For a standard LP, it often means a small increase in that constraint’s right-hand side does not improve the objective at the current solution. It does not mean the resource is worthless in every scenario; a larger change or a different model may alter its value.

Can I trust the shadow price outside the allowable range?
No. The report’s marginal estimate is supported only within its stated range, with other inputs held fixed. Beyond that range, the optimal basis may change, so update the model and solve it again.

Does a sensitivity report work with integer or binary decisions?
Not in the usual LP sensitivity sense. Integer and binary decisions make the problem different from a continuous linear program, so use Solver’s appropriate solution reports and test revised assumptions by solving the model again.

Related

Try it: the Date Difference Calculator: days, months, years and weekdays between two dates, with the DATEDIF formulas.

More in How-to basics

Sources: Function behaviour and menu paths: Microsoft Support documentation for Excel. Method & sources

Spotted an error in this guide? Let us know.