Is it worth it, and when does it pay back?

How the calculation works

Year 0 carries the upfront cost. Each later year carries the running cost and, from the first benefit year on, the benefit that the program actually causes. Every amount is discounted back to year 0, so a dollar in year 5 counts for less than a dollar today.

• Net present value is discounted benefits minus discounted costs. Above zero, the program returns more than it costs at your discount rate.
• Benefit per dollar is discounted benefits divided by discounted costs. Public bodies often call this the benefit-cost ratio.
• Return on investment is net present value divided by discounted costs.
• Payback is the year in which cumulative discounted net value turns positive. With a discount rate of 0 it is simple payback.
• Break-even benefit is the yearly benefit at which net present value is exactly zero. If you are unsure of the benefit, this is often the more useful number. Ask whether the program can plausibly deliver at least this much.

The input most people skip

"Would have happened anyway" is the share of the benefit you would have seen without the program. Staff who would have learned the skill on the job. Customers who would have bought without the campaign. Clients who would have recovered on their own. Evaluators call it the counterfactual, or deadweight in social return on investment work.

It often decides the answer. A program that looks like a 3-to-1 return can drop below break-even once half the benefit turns out to have happened anyway. Estimating that share well is the hard part of any evaluation. Our article on difference-in-differences shows one way to measure it with a comparison group.

What it leaves out

• Uncertainty. The calculator uses one value for each input. Try a low and a high case for the benefit and the share that would have happened anyway, and see whether the answer changes sign.
• Benefits that are hard to price. Health, wellbeing or reputation gains need a money value first, or a separate cost-effectiveness measure.
• Growth or decline over time. Benefits and costs are flat each year. Real programs ramp up and wear off.
• Tax, financing and inflation. Use real amounts and a real discount rate, or nominal for both.

The examples are illustrative round numbers, not data from any organization. For the discount rate, use your organization's own. Many governments and funders publish one for appraisals.

Related: What is an estimand, and why does your program evaluation need one? and our program and policy evaluation service.