Sensitivity analysis: see which assumptions matter
Sensitivity analysis asks how a result changes when an assumption changes. It helps direct attention to the inputs worth checking and the choices worth examining.
Does a higher price improve the result if demand responds?
This is a playable answer made in flatland desk. Move its measured settings to explore the result. The assumptions are fictional.
Goal · 2027-01 through 2027-12
Operating profit reaches at least $60,000 over one year
Starting result
$55,500
On a narrow screen, scroll the answer sideways to see its full width. Use Tab to reach its settings and the arrow keys to change them.
Open the answer in its own tab · Download the model for desk
These results were measured when the answer was made. The page does not run new simulations. To change formulas or ranges, open the model in desk.
Start with one change
What-if analysis compares a specific change with the starting assumptions. In this model, changing price also changes modeled demand. Follow both effects before judging the result.
Move the price setting in the playable answer and compare the solved result with the goal’s odds. A higher result at the starting demand assumption need not mean better odds under uncertain demand.
Separate choices from conditions
A lever is a choice within the model’s feasible range. A force describes something uncertain that the person does not directly control. Sensitivity to a force is a reason to investigate or prepare; it is not an instruction to set reality to a convenient value.
flatland desk distinguishes the drivers that affect the result and the levers that improve the goal’s odds. The ranking belongs to this model, its ranges, and its goal. It is not a universal ranking of what matters.
Turn the result into a next step
If the decision depends heavily on demand response, gather evidence about demand before changing the price. If two choices have similar results, check their costs and practical constraints outside the model.
This example uses fictional assumptions. Its purpose is to make the reasoning inspectable. It does not recommend a price for a real business.
Build your own model
flatland desk by Flatland Labs makes a quantitative model’s structure, behavior over time and possible futures visible.
Downloading and local use need no account. A plan includes the built-in agent and the handoff.
Why flatland exists, and where it’s going: Escape.