Home · Posts
Quantifying Failing Fast
March 7, 2011

When you're choosing between opportunities to pursue, and there is any uncertainty in their outcomes, be sure to consider how long each possible outcome would take. Opportunities that would fail fast are more likely to be worth doing. People often neglect this effect, perhaps because its magnitude can be surprising.

Suppose you're weighing job offers from two companies. One's well-established; you're guaranteed a good return on your time. The second is an unproven startup that's giving you significant equity. You stand to earn a lot if things go well, but it could also crash and burn.

There are many factors to consider when making a choice like this one. Most of them are intuitive. For example, almost everyone would think about the risk premium the startup has to provide, even if they don't know that term. But for some reason, many people would tend to calculate the value of each opportunity over a falsely unified four-year time frame. This totally misses the additional value that comes from failing fast, which can lead to the wrong decision.

Consider this simple model, which I've made extreme in order to demonstrate the principle.

  • At the established company, your compensation is worth a reliable $100k per year.
  • The startup has a nine-in-ten chance of total failure: you get nothing. Importantly, though, this case only takes one year — the assumptions were wrong, Google does it, the team implodes, whatever. In the other case (one in ten) it achieves glorious liftoff, and you'll end up after four years with a cool $4M.

In the naive assessment, you think "the startup opportunity is worth $400k on average, but if I spend those four years at the other company, I'd have made the same amount". It would seem you should simply go with the less risky one. But what you really want to think about is how much you'll make per year. This figure incorporates the fact that you do get something when the startup fails: three years to try other things! From the broader perspective of your career, that makes value per year the right metric, not value per opportunity. (Notice how an investor's incentives would be different.)

An easy trick for calculating the value per year of the startup opportunity is to play out all the scenarios: nine times, we spend a year and make nothing, and then one time we spend four years and make $4M. In total, that's $4M divided by 13 years, for an average earnings of $307k per year. That's a lot more than the established company, even though the naive assessment made them look equivalent. (Of course, risk and expected utility are still important. I'm just pointing out one important factor to consider.) Over a lifetime of making such choices — including, more broadly, ones that aren't quantifiable — the difference can add up.

Thanks to Aston Motes.