Three Cardinal Sins of Investing: How Artificial Intelligence Can Help Identify Founder Fit

What do Amazon, Alibaba, AirBnB, Bentley Systems, Cisco, DataDog, PagerDuty, Peloton, Pinterest, Robinhood, SquareSpace, Uber, Udemy and UIPath have in common?

In Secrets of Sand Hill Road, Scott Kupor described the three cardinal sins of venture capital: (1) investing in the right team but wrong market; (2) investing in the right market but wrong team; and (3) not investing in the right market with the right team. Of these three, investing in the right team is least problematic as an A team can sometimes make a B plan work or pivot to a better place, but a B team will fumble an A plan every time. Management misjudgments are always painful. Opportunities to invest in markets that produce large, outsized returns are rare so backing the right team in the right market is essential to create an outperforming fund.

Yet investors routinely pass on high potential teams pursuing large, disruptive opportunities. Amazon, AirBnB, Bentley Systems, Cisco, DataDog, PagerDuty, Peloton, Pinterest, Robinhood, SquareSpace, Uber, Udemy and UIPath are publicly traded unicorns that together are valued at over $2.5 trillion. Yet each of these technology firms had difficulty raising their first round of funding. Jeff Bezos spent over a year raising Amazon’s first financing round. AirBnB, Pinterest, Robinhood and Udemy founders claimed that over 100 venture firms initially passed on their startups. John Foley said raising the first three rounds for Peloton was ‘bone crushing.’ Y Combinator rejected Alex Solomon, founder of PagerDuty, four times before admitting him on the fifth try after he had achieved product market fit. Bentley Systems, SquareSpace and UIPath founders bootstrapped their businesses for over fifty years cumulatively before raising financing.

These iconic companies generated exceptional returns, yet early investor rejection was the norm. Nearly half of the disruptive technology firms that had unicorn IPOs in the past five years claimed to have difficulty raising funding. As illustrated in Table 1, hundreds of top U.S. venture firms committed the third cardinal sin of investing in these businesses: they made an error of omission and passed on a startup with the right team going after the right market.

Table 1: The Three Cardinal Sins of Investing

Successful venture investors use pattern recognition to make investment decisions. We meet thousands of entrepreneurs and review hundreds of funding pitches annually. Investors reliably apply pattern recognition for startups promoting incremental improvements as founders generally have prior industry experience. Pattern recognition for disruptive startups is harder as founders typically come from outside the industry and often lack telltale markers of success. Yet disruptive innovation tends to produce the largest outcomes, so pattern matching is least attuned for ventures that matter most.

Founder Opportunity Fit: Pattern Recognition Challenges

Two challenges may explain our low hit rate in assessing founders of disruptive startups: (1) these entrepreneurs are a diverse lot that defy pattern recognition; or (2) investors misread entrepreneurial talent who lack the pedigree of founders they typically back. Research suggests that both factors apply.

Successful entrepreneurs are a diverse lot and rarely conform to central casting. Few investors would wager on the following profiles:

· Adopted child. College dropout. Fired from first company. Unkempt and rarely showered. Cheated his cofounder out of 80% of his annual bonus.

· Failed college entrance exam twice due to weakness in math. Worked at Kentucky Fried Chicken after rejection for 31 other jobs after college. Went into teaching after his first company failed.

Those who declined to invest passed on Steve Jobs at Apple and Jack Ma at Alibaba. Masayoshi San of Softbank invested in Alibaba claiming that Jack Ma had the “scent of an entrepreneur.” Sounds like Steve could have used some of Jack’s cologne during his long intervals between showers!

If there is a scent of an entrepreneur, then investors need to adjust their olfactory senses. Investors are susceptible to the liking bias: they tend to invest in people with shared traits and backgrounds. Since funds and startups require different skills, investors may be drawn to the wrong sorts of folks.

Biases tend to compound in hot markets when overconfidence and the Fear of Missing Out (FOMO) prevail. Charlie Munger warns of a “Lalapalooza Effect” when multiple biases converge. Receding tides show who is swimming naked as we observed with now bankrupt FTX where funds invested $1.8 billion without any board oversight to catch onto the cryptocurrency craze. Sam Bankman Fried had all the right credentials but lacked operating experience and would have benefited from some oversight.

NGP Capital has long evaluated management teams of prospective investments using a checklist of desirable attributes. These assessments provoke thoughtful discussion on founders, yet scores often coalesce within a narrow range and are rarely illuminating. Our experience mirrors those of investors and corporations alike: management assessment systems are highly subjective, risk reinforcing existing biases, and are readily reverse engineered for desired outcome.

Artificial Intelligence: How Computers Can Help Identify Founder Opportunity Fit

Great challenges present great opportunities. Leadership is the most important factor for startup success. As Jim Collins noted in Good to Great: “First who … then what.” Conversely, Peter Thiel from Founders Fund observed: “A startup messed up at its foundation cannot be fixed.”

Venture capital thrives on solving seemingly intractable problems. Much as stock brokerage firms invest billions to incrementally improve arbitrage trading systems, venture capital is highly incented to develop systems that can identify and assess high potential talent.

Artificial intelligence holds much promise for the venture industry not only as a target but also a tool for investment. Computers generally perform better than humans at pattern recognition. Machine learning has uncovered surprising efficiencies in many other fields, including energy and utilities, transportation, logistics, manufacturing, retail, financial services, agriculture and poker.[1]

Like snowflakes, effective entrepreneurs come in all shapes and sizes, but they tend to converge on a set of common characteristics. Table 2 contains a list of commonly identified characteristics of successful entrepreneurs. While your preferred traits may differ, we may agree that many traits are both subjective and nuanced: a high score is preferred for some while a medium or low score is better on other measures. The nuanced, subjective nature of management evaluations suggests that we may benefit from an informed third-party assessment.

Table 2: Selected Startup Leadership Traits

In venture capital, as in many fields, augmented intelligence is more effective than artificial intelligence. Machine learning systems that assess management teams should augment, not replace, investment judgment.

The column on the far right of Table 2 shows how human judgment and artificial intelligence may complement each other. At least for the near future, veteran investors may outperform computers in evaluating nuanced traits that are not readily measurable such as those in red. Machine learning may have an advantage in areas that are more quantifiable and where larger data sets could provide broader assessments such as those in green. Equally important are blue areas where machine learning could help reduce bias and augment human intelligence.

Several firms are testing machine learning systems to address this opportunity. In the past week, ZipRecruiter announced a new service to automatically schedule interviews with candidates who match job descriptions using artificial intelligence. Harmonic.ai identifies high potential companies using market and founder profiles provided by investors. CrystalKnows.com is addressing more nuanced traits and, based on a cursory test, offers useful insights.

These firms are wise to begin with artificial intelligence solutions that focus more on discovery than evaluation and offer solutions that augment human intelligence and enhance efficiency in the laborious discovery process. Discovery solutions may also better comply with privacy regulations such as General Data Protection Regulation (GDPR) in Europe.

Venture firms may also start using machine learning to identify high potential talent and augment networking efforts. Finding ways to engage investors and elicit feedback is essential to train machine learning systems. Eventually, improving systems may coincide with investor confidence as they use the tool and incorporate feedback into investment decisions. Any tool that helps reduce the three cardinal sins of investing would be welcome.

Three Cardinal Sins of Investing: Related Concepts

The Three Cardinal Sins of Investing involve misevaluations of founders and markets and highlight the importance of correctly assessing Founder Opportunity Fit and Product Market Fit. Investing in the right market but wrong team involves an Opportunity Cost if it precludes an investor from making future investments in this high potential market. Investors risk Adverse Selection if they consistently apply suboptimal methods for founder assessments.

Many factors complicate startup founder evaluations. Radical or 0 to 1 Innovation often emerges from outsiders, so industry experience is often a negative indicator for disruptive startups. Information Asymmetry is most pronounced in early-stage investments, so investors may routinely pass if they do not understand the space well. Cognitive biases, especially Like Bias or Associative Bias, may be more prevalent in early stage investing where there are few objective guideposts.

The benefits of correctly assessing Founder Opportunity Fit are significant. Power Law Distributions where 80–90% of venture returns come from 10–20% of invested capital make misevaluations especially costly but also justify technology investments that can even marginally improve success rates. Investing in a great founding team increases Margin of Safety as they can course correct if their initial plan proves unsuccessful. Systems that consistently improve founder assessments offer a Sustainable Competitive Advantage for their investors as such systems are neither visible nor readily replicable by other investors.

[1] See On the Edge by Nate Silver; Thinking in Bets: Making Smarter Decisions When You Don’t Have all the Facts by Anne Duke; and How to Decide: Simple Tools for Making Better Decisions by Anne Duke.