NPV vs IRR vs payback: what each measure says about an investment, why they disagree, and which one decides when you can fund only one project.
By Sergey BruhPublished 6 min read
A free course for PMs and marketers: short lessons, real product data and exercises with instant feedback.
Start the free courseNPV vs IRR vs payback, in one line: decide with NPV, show IRR and payback next to it. Net present value (NPV) tells you how much value an investment adds in today's money at your required rate. The internal rate of return (IRR) is the rate at which that NPV would be zero: a handy summary, but blind to the size of the bet. The payback period says how long your money is at risk and ignores everything after. When the three disagree, and they often do, NPV wins. Below: a worked example with three product bets where each measure crowns a different winner.
is the cash flow in year , with the investment as a negative number. Discount cash, never accounting profit: depreciation is not money leaving the bank, and a prepayment is money arriving before the profit does.
Klasna, an online tutoring school, has one product team and three ideas for next year. Each row is the extra contribution minus extra running costs, in UAH; year 0 is the build, paid now, later years arrive at year end. Klasna's required rate is 25% a year.
Payback. The app gets 400,000 back in year 1 and needs 200,000 of year 2's 300,000: years. The mock-exam mode is at −900,000 after year 1 and −300,000 after year 2, then needs 300,000 of year 3's 800,000: years. The reminders: years.
NPV at 25%. The discount factors are , , and :
The app: . The reminders: .
IRR. Solved numerically (a spreadsheet's IRR function does it): mock exams 31.26%, app 11.51%, reminders 34.58%.
Each measure picks a different winner. Payback favors the reminders and ranks the app above the mock exams, yet the app destroys 88,000 of value: it returns 700,000 for 600,000, just not fast enough to earn 25%. IRR favors the reminders too. Only NPV picks the mock-exam mode, and it's right: if the team can build one thing, 161,280 of added value beats 20,800. A high rate on a small sum is still a small sum.
IRR is a percentage, so it can't see scale: 34.58% on 200,000 adds less than 31.26% on 1,200,000. It also favors bets that return cash early. For mutually exclusive options (one team, one budget, one slot), rank by NPV. IRR is safe as a yes/no check against your required rate for a single, ordinary project; it's unsafe as a ranking tool.
IRR has two more quirks. When cash flows change sign more than once (a big refit in year 3, say), there can be several IRRs or none. And it quietly assumes interim cash can be reinvested at the IRR itself, which is optimistic for a high-IRR bet.
Payback can use today's money too. Add discounted cash flows instead: for the mock exams the running total is −960,000, −576,000, −166,400, then +161,280, so discounted payback is years instead of 2.38. More honest about waiting, but still blind to whatever comes after the payback point. Use it when cash is tight and you need to know how long the money is out, not to choose between bets.
NPV, for decisions. It measures value in money, so it handles bets of different size and timing. IRR is useful alongside it as a quick sense of the return.
There is no universal number. Compare it with how long your cash can wait and how long the advantage lasts: a feature competitors copy in two years needs to pay back well before that.
Your cost of capital for an average-risk project, plus a premium for riskier ones such as a new market. A company's loan rate alone is too low, because loans fund only part of it.
Yes. The mock-exam mode pays back in 2.38 years, later than the app, yet adds 161,280 while the app loses 88,000.
Module 6 of the free course Finance for Product People builds this step by step on Klasna's numbers: start with the time value of money, then NPV, IRR and payback for a product bet, and finish with a real choice: build our own video classroom or keep paying per minute.
