On Failureship Bias
One of the biggest mistakes we make in Silicon Valley is generalizing reasons why startups fail. Trying to generalize reasons for failure across companies has two big problems. The first is that it creates an odd form of survivorship bias I call failureship bias. We fixate on common and visible failures. This leads to overlooking incredible companies. The second is that the attribution of failure is difficult. This creates mistaken conclusions about what future startups should avoid. The nature of venture-backed startups is that their returns follow a power law. Successful startup building and investing is about exception finding, not about avoiding failure.
The Perils of Failureship Bias
The study of startup failure overemphasizes the spectacular or the repeated. The failure of Theranos had a best seller written about it. Should we learn more from Theranos’ failure than any other startup? At Y Combinator, they highlight the two most common reasons for failure: cofounder issues and not finding product-market fit. These causes of failure have lessons. Founders should do their best to avoid these failure modes. But the ecosystem shouldn't. If we blacklisted companies where founders fired each other, we wouldn't have Twitter. There are exceptions to every red flag a startup could have. Shouldn't one avoid companies that brought in an adult CEO reporting to the founders? Surely, the dysfunctional executive arrangement at Google (who was the CEO?) was a driving cause of failure. How about married founders? Eventbrite seemed to do okay. Companies that don’t have a great product? Craigslist. Situations where a founder had started two companies simultaneously? SpaceX and Tesla (this one is a bit personal). Upside down unit economics? Uber, Lyft, DoorDash (ask an early stage investor if they're happy). No revenue model or monetization? Google and Facebook. Every successful Silicon Valley startup is an exception to a rule.
What are some of the things I've done despite them being red flags? I've started two companies at once. I've parted ways with a cofounder. I've brought in a "professional" CEO very early. I've spent a large chunk of the money we raised on acquiring another startup before we had a lick of revenue. And these decisions, I believe, were the correct ones. We will see how my entrepreneurial endeavors shake out, but the point stands. Every company requires its own exception handling and has to place its bets on what those things are.
Startups are about finding exceptions. There's a perverse logic to finding reasons not to invest based on historical patterns from heterogeneous companies. Because the financial incentives of VC's drive the strategies of every company they back, the strategies of venture-backed companies should be dependent on the question, "What could go right?" This is scary as an entrepreneur. You don't have a portfolio to distribute your risk. But this is the game you signed up for. The way to give yourself a portfolio is to increase your number of shots on goal. Most of what you do as a startup is failing and learning what not to do. This is the hard part of being iterative as a startup and it is why belief matters. Stay committed to your vision and try more paths to get there. Don't fall victim to the homogeneity of Silicon Valley's thinking.
The Challenges of Attribution and Generalization
Secondly, it is impossible to attribute the true reason for failure. Consider the two most common failure reasons from YC: cofounder fights and product-market fit (PMF). When cofounders fight, is it because the founders hadn't worked together in the past? Or was it that failing causes fights and they were struggling to find PMF? When a startup can't find PMF, it’s often because the team is bad at building/designing products or they chose a bad idea. This expands to other, less common, reasons. When customer acquisition is too expensive, it’s often a sign the market is too small or the team never hired the right growth or channel acquisition talent. Attribution focuses on the most proximate reason for failure, not the underlying one. It's something like a greedy algorithm for understanding failure causes.
To then apply these learnings to a different startup is an exercise in futility. Consider those who insist that startups should raise less versus those who insist more capital is better. These are broad generalizations derived from prior experiences. You know how happy Uber is that they raised too much? Very. You know how sad numerous companies are that they raised too much? Very. But how many of the companies that "raised too much" instead lacked focus and execution? Some companies need to raise too much, others need to raise less. Most venture capitalists don't have enough time to attribute failure. When you sit on 20+ boards, you can’t deduce nuanced learnings from one. When you can't even get the base reason right, applying those learnings to a new company is an uphill challenge.
With startup strategy, people have far too much certainty in an exercise that is inherently uncertain.
Exception Finding and Loss Aversion
Loss aversion creates pervasive avoidance of failure. The nature of human psychology dictates that each spectacular failure hurts far more than each success brings joy. Most venture capitalists cannot overcome the psychological reality of loss aversion and it drives them to reduce risk across the system. Once a company reaches a certain stage, VC's spend less time on them, making them less top of mind and further exacerbating this problem.
The psychology of loss aversion is a primary driver of missed opportunities. One example I credit to Josh Kopelman of First Round Capital is the story of YouTube. YouTube was not the first, nor the second, nor the third video streaming company. As Josh describes it, no less than 20 video streaming companies came before YouTube. Each promised an incremental improvement (no plugins, faster buffering, etc.) over the previous ones. Imagine being the first to invest in a video streaming company on the web. The company fails. You're a high conviction person, so when the 6th company comes around, you invest again. It fails. You then see YouTube. It would be insane to invest in a third company of the same type. You pass. You miss out on a billion dollar exit. What's the tragedy of this? You were right! You had the exact vision! And YouTube is no exceptional story. Google was the 24th search engine. Facebook was the at least the dozenth social network (props to Mark Pincus who started a failed social network but still backed Facebook). A company will come around one day doing what Theranos tried to and many will pass. Learning too much from the past given our cognitive biases is dangerous. Consider the math of it: if your prior probability of success is less than 1%, failure should not revolutionize the prior.
The technology industry is ripe full of examples of successful repeat efforts. Spectacular failures amplify failureship bias to even greater levels. Think of the opposite, how many times do you hear stories about promising startups that failed because the market was too small or fractured? People tend to forget those companies. Companies whose growth stalls out at a market cap of $100m aren't spectacular failures but they are venture losses.
So what?
So why does this matter for entrepreneurs? Fundability impacts your strategy. The incentives of your venture investors impact your strategy. Loss aversion rules supreme. This is not advice to mimic past failed strategies but you should align yourself with people who are not loss averse. Avoiding loss aversion is a market inefficiency. You should always have a hypothesis of why you will succeed where others failed, but don't be afraid because investors have scars.
