Two projects land on your desk. One promises a 40% return; the other promises 18%. The easy call is the 40%, and it can be the wrong one. The reason is that "return" is not one number. It is at least five, each answering a slightly different question, and a leader who can't tell them apart is at the mercy of whoever built the spreadsheet.

The quick version

  • NPV answers "how much value, in today's money, does this create?" It is the tie-breaker. When the others disagree, trust NPV.
  • IRR is the project's built-in percentage return. Intuitive and popular, but it flatters small, fast projects and can mislead on big ones.
  • Payback asks "when do I get my money back?" A useful risk gut-check, but it ignores everything that happens after, and the time value of money.
  • ROI is a rough overall multiple; ROIC is the disciplined cousin that asks whether your returns actually beat your cost of capital. Above that line you build value; below it, growth burns cash.

The idea in depth

Every one of these tools exists to answer the same underlying question, is this a good use of money we could spend elsewhere?, and they differ only in what they choose to ignore. Understanding what each one drops is the whole skill.

NPV: the one that actually answers the question

Net present value takes every future cash flow a project will produce, discounts each back to today's money at your cost of capital, and sums them. Positive NPV means the project creates value over and above what that money could earn elsewhere at the same risk. The decision rule is brutally simple: take every project with a positive NPV; among mutually exclusive options, take the highest.

This isn't a fringe academic preference. In their landmark survey of 392 chief financial officers, Graham & Harvey, Journal of Financial Economics, 2001, roughly three-quarters of CFOs reported always or almost always using NPV, putting it neck-and-neck with IRR as the dominant capital-budgeting method. The standard finance text, Brealey, Myers & Allen's Principles of Corporate Finance, treats NPV as the benchmark every other rule is measured against, precisely because it is the only one denominated in the thing shareholders care about: money, today.

So the move is: make NPV your default decision rule and your referee. When two metrics point different ways, the project with the higher NPV is the one that adds more value, full stop.

The honest limitation: NPV is only as good as its inputs. It collapses a messy future into two assumptions, the cash-flow forecast and the discount rate, and small changes in either swing the answer hard. A confident NPV built on a hopeful forecast is just a hope with a decimal point. Always pressure-test the assumptions, not the arithmetic.

IRR and payback: the popular numbers that mislead

IRR, the internal rate of return, is the discount rate at which a project's NPV equals exactly zero. People love it because it comes out as a clean percentage you can compare to a hurdle rate: "23% beats our 12% threshold, approve it." In the same Graham & Harvey survey it was the single most-cited method. But that tidiness hides three traps. IRR implicitly assumes you can reinvest the project's cash flows at the IRR itself, often unrealistically high. It can produce multiple answers when cash flows flip from negative to positive more than once. And it systematically favours small, fast projects: a 50% return on $10,000 looks better than a 20% return on $10 million, even though the second creates vastly more value. This is why finance texts are near-unanimous that when IRR and NPV disagree on mutually exclusive projects, you follow NPV.

Payback, how long until cumulative cash flows repay the initial outlay, is the simplest of all, and the most openly flawed. It ignores the time value of money entirely, and it goes blind the moment your money is back: a project that pays off in two years then collapses scores better than one that pays off in three and prints cash for a decade.

So the move is: keep IRR and payback as supporting evidence, never the verdict. Use IRR to sense the cushion above your hurdle rate, and payback as a fast risk read, short payback means less time exposed to a forecast going wrong, which genuinely matters in volatile or cash-tight settings.

The honest limitation: their popularity is the danger. Because IRR and payback are intuitive, they dominate the room, and a leader who lets the cleanest-sounding number win will, often enough, fund the smaller, shorter, lower-value option.

flowchart TD
    A(["A project lands on your desk"]) --> B{"Is NPV positive at our cost of capital?"}
    B -->|No| C(["Reject, it destroys value"])
    B -->|Yes| D{"Choosing between rival projects?"}
    D -->|No| E(["Fund it"])
    D -->|Yes| F(["Pick the highest NPV; use IRR & payback as tie-break colour, not the verdict"])
A simple decision order: NPV decides, the others inform. Leaders Loop

ROI and ROIC: the difference between "we made money" and "we created value"

ROI, return on investment, is the back-of-envelope ratio: gain over cost. It is everywhere because it is easy, and it is dangerous for the same reason. A 30% ROI tells you nothing about over how many years, or whether 30% even cleared the bar of what that capital costs you.

ROIC, return on invested capital, fixes the part that matters most. It compares the operating profit a business earns to the capital tied up generating it, and then sets that against the cost of that capital. McKinsey's corporate-finance team makes the consequence stark in their analysis of what drives shareholder returns: a company creates value only when its ROIC exceeds its cost of capital. The same test anchors Valuation, the long-running McKinsey textbook by Tim Koller, Marc Goedhart and David Wessels. Below that line, growth doesn't help, it actively destroys value, because every extra dollar invested returns less than it cost to raise. Warren Buffett has made the same point for decades in plainer English; in his 2007 Berkshire Hathaway shareholder letter he held up See's Candies as the ideal business precisely because it threw off enormous profit on a tiny amount of invested capital, a ferociously high return on capital, not merely a big absolute profit.

So the move is: stop celebrating growth and ROI in isolation. Before you scale anything, ask whether its ROIC clears your cost of capital. If it does, pour fuel on it. If it doesn't, growing it faster just loses money faster, fix the returns first or don't grow it at all.

The honest limitation: "invested capital" and "cost of capital" are both estimates with real judgement baked in, how you treat leases, goodwill, intangibles and R&D can move ROIC several points. It is a discipline and a direction-finder, not a precision instrument. Use it to ask the right question, not to win an argument to two decimal places.

flowchart LR
    A(["ROIC vs cost of capital"]) --> B{"Does ROIC beat the cost of capital?"}
    B -->|"Yes, spread is positive"| C(["Growth creates value, invest to scale"])
    B -->|"No, spread is negative"| D(["Growth destroys value, fix returns first"])
The value-creation test: it is the spread over cost of capital, not the raw return, that decides whether growth helps. Leaders Loop

A worked example

All figures below are illustrative, chosen for round arithmetic, not a real company.

You can fund exactly one of two projects. Your cost of capital is 10%.

  • Project Quick: spend $100k today, get back $130k in one year. IRR = 30%. Payback ≈ 0.8 years. NPV ≈ $130k ÷ 1.10 − $100k ≈ +$18k.
  • Project Big: spend $1,000,000 today, get back $1,250,000 in one year. IRR = 25%. Payback ≈ 0.8 years. NPV ≈ $1.25m ÷ 1.10 − $1m ≈ +$136k.

Rank by IRR and Project Quick wins, 30% to 25%. Rank by NPV and Project Big wins by a mile, it creates roughly $136k of value versus $18k. Same hurdle, same one-year horizon, opposite verdicts. The IRR is "right" that Quick is more efficient per dollar; it is silent on the fact that you have far more dollars to deploy in Big and they all clear the bar. Because these are mutually exclusive, you take the bigger NPV. The lesson in one line: a high percentage on a small base is a smaller win than a decent percentage on a large one, and only NPV tells you that directly.

Frequently asked questions

If NPV is best, why does everyone still quote IRR?

Because a percentage is easier to carry into a meeting than a dollar figure that depends on the discount rate you assumed. IRR needs no stated cost of capital to compute, and "beats our 12% hurdle" is a sentence anyone can act on. It is a fine headline, just not the final word. When IRR and NPV agree (they usually do), quote IRR for the room and keep NPV as your private referee.

What discount rate should I use?

For most corporate investments, your cost of capital, usually the weighted average cost of capital (WACC), the blended cost of your debt and equity. The riskier and more uncertain the project, the higher the rate it should have to clear. Getting this number roughly right matters more than getting NPV precisely right, which is why it has its own Toolkit entry.

Is payback ever the right tool?

As a primary decision rule, almost never. As a risk filter, often. In a cash-strapped business, or a fast-moving market where forecasts beyond three years are fiction, a short payback genuinely lowers your exposure to being wrong. Treat it as a constraint ("nothing over a four-year payback") layered on top of NPV, not as the ranking itself.

What's the real difference between ROI and ROIC?

ROI is a loose ratio of gain to cost with no agreed definition and no benchmark. ROIC is disciplined: profit over the capital actually tied up, judged against the cost of that capital. ROI tells you a project made money. ROIC tells you whether it created value, whether the money it made beat what the money cost. Only the second is a basis for deciding where to grow.

Which one number should I ask for first?

For a single go/no-go decision, ask for NPV at your cost of capital. For deciding where to invest more across a portfolio of businesses, ask for ROIC against cost of capital. Those two cover most of what a leader actually needs; IRR and payback are the colour around them.

Related in the Toolkit

Where to go next