We are excited to announce that Mattermark has re-launched as an independent company! Learn More →
Back to All Articles  |  Editorial

Why IRR Is The Most Important VC Performance Metric (That’s Also Nearly Impossible To Use)

tl;dr: IRR is theoretically the best way to measure the performance of a venture capital firm. In practice, it’s far harder to use effectively than you might expect.


As Jason Rowley wrote yesterday in “Everything You’ve Ever Wanted To Know About VC Returns (But Were Afraid to Ask),” most measures of venture capital returns are squishy.

That’s especially true for the first three to five years of a fund, a time window of key importance, as firms often raise new funds every one to four years; therefore, potential limited partners (LPs) often base investment decisions on that previous fund’s more nascent results.

Rowley contends that net present value (NPV) is the best way to measure venture returns. However, internal rate of return (IRR), if measured correctly, is the best way to evaluate venture returns. Let me explain.

Measure This

Measuring returns based on realized dollars returned to date will usually lead to a lower measured return until the final close of the fund (when the fund exits its last investment) compared to other methods. Most startups that fail often fail fast, so a fund’s early realized returns are often negative.

For example, say Fund ABC invests $10 million in 10 startups. Five exit in the next year, two of which eventually yield nothing, and the other three return $5 million, $8 million, and $12 million. Realized returns would total -50 percent ($25 million returned from $50 million invested).

Meanwhile, the next unicorn the fund is invested in won’t return any money to investors for a long time. So your returns start low, go negative, and then go up–in theory. 

It’s called the J-Curve.

Changing tack, trying to measure value using NPV requires forecasting several years’ cash flows. This is helpful in private equity, where companies are already often generating predictable cash flows. However, it’s harder for younger companies.

Try to extrapolate cash flows out 10 years for a pre-revenue startup. No seriously, try it. Even if you know the market and company well, your answer will be way off. The same is true for a startup that is currently in growth mode and sporting largely negative cash flows.

If you and I were good at forecasting cash flows, NPV would be the best measuring tool for venture capital performance. Unfortunately, humans are not, and venture capitalists tend to be overly optimistic, introducing an upward bias to cash flow estimates.

Turning to another method, return on investment (ROI) is a great value metric. It is a mostly objective measure when gains are realized, but ROI doesn’t take into account the time value of money.

Let’s imagine a few scenarios.

  • You can invest in Alexandra’s “Quick Flip Fund” that gives you a 1.3x return on your money, but it only takes a year for her to successfully exit investments. Tour return is 30 percent1.
  • You can invest in Alexia’s “Long Term Fund,” which will get you a 5x return on your investment in year 7.
  • However, if Alexandra can get a 30 percent IRR every year (and you fully reinvest returned capital), after 7 years, the combined ROI is 6.3x.

Which would you rather have? While ROI is great in some aspects, it doesn’t take into account that a dollar today is worth more than a dollar 5 years from now.

Moving past both NPV and ROI, I’d argue IRR is the better measure of venture returns — when it is measured correctly and effectivelyEssentially, IRR enables the LPsto measure their venture returns against other asset classes in a comparable manner.

However, there are difficulties to the metric. For example realized IRR is by far the best objective measure of venture returns — but only after the fact, which makes IRR nearly useless when benchmarking the performance of funds to decide who is better at investing today. After a firm closes down a fund in year 10 after its cycle is complete, most, if not all, of the firm’s bets will have been placed 6-10 years ago. 

Realized IRR is very important for reporting how well an LP’s investment in the fund performed. But using realized IRRs is looking backwards. You can only measure it when everything is said and done. It’s like ranking a sports team based on their performance 7 years ago.

Extending the analogy, the Washington Nationals currently have the second-best record in the National League. Back in 2008 they had the worst record in baseball.

Estimating The Unrealized

As noted above, venture capitalists tend to be optimistic and overestimate valuations. Combined with even a touch of financial magic (like pushing markdowns into the next year by a day), IRR is mere SWAG, or a Serious Wild-Ass Guess, at best. At worst, it’s a heavily mutable number.

With median time to exit at almost 7 years for IPOs, trying to assess the value of what something as volatile as a startup will be worth that far in advance is a fool’s errand.

Even trying to estimate what something is worth right now is nigh on impossible (just look at all the quarterly valuation perturbations for venture investments held by top mutual funds like Fidelity and T-Rowe Price). And those same mutual funds often can’t even agree on what those companies are worth.

So if large mutual fund companies — with $100 billion plus in assets under management and a long history of valuing companies — can’t come up with a valuation that matches the market, why in the world would we believe venture firms can do it effectively?

It’s The Best And Worst, But It’s Only The Best When It’s The Worst

In theory, IRR is the best measure of venture returns.

But, in practice, unrealized IRR is prone to human judgment errors and system-gaming. And using realized IRR is tough because it’s a trailing metric.

Hell, a16z was only founded in 2009. How are you going to compare 7 years of returns to Sequoia’s almost 50 years of returns?

So how should we measure venture firm’s returns? I don’t have a perfect answer; it is such an illiquid and unique asset class. My best guess would be to take the average of both NPV and overall — including unrealized — IRR. 

IRR, when it is measured with care and as little bias as possible, is probably the best measure. But in the end, LPs have to trust their venture funds have their best interests at heart.

After all, VCs always behave.

Scott is the head of US Sales and Marketing at SmartCat.io, a big data consultancy. He has worked in various startup roles, as a VC, and as an Equity Research Analyst. If he actually paid his dues, he would also hold the CFA designation.

  1. This example discounts for fees, and so forth for the sake of simplicity.

Join thousands of business professionals reading the Mattermark Daily newsletter. A daily digest of timely, must-read posts by investors and operators.

© Mattermark 2024. Sources: Mattermark Research, Crunchbase, AngelList.