Jockeys, groundwater and other innovation myths

I’m a data scientist who has spent nearly a decade trudging through quantitative data about what predicts the success or failure of innovations. Yes, really. I’ve done this because I genuinely care about innovators and want more of them to succeed. I’m also a venture capitalist, so my interest is applied as well as theoretical. In this spirit, here are counter-intuitive lessons whispered to me by data that, apparently, irritate a lot of people. Enjoy!

myth-reality

It’s all about the team… nope.

I’ve written about this extensively, as have gobs of academics for decades, but it still seems to touch a nerve. Vast quantitative studies about the impact of leadership on firm performance have found either no statistically significant impact, or a small one at best. My research has basically found the same thing, which is – a bad team can kill any business, but a great team of rockstars can only increase your odds of success by around 12%. So yes, there is a difference, but it doesn’t account for 88% of the whole picture.

Statistically speaking, success seems to be more about avoiding bad teams rather than staffing up with rockstars. Past success doesn’t immunize people from future failure. Thomas Edison himself failed far more than he succeeded… Edison for cripes sake! It doesn’t get more rockstar than that. Anyone who says “bet on the jockey, not the horse” has never tried to win the Kentucky Derby on a three-legged mule.

Get an effective team, avoid a bad team, and otherwise don’t stress over it.

It’s all about the investors… nope.

Venture capitalists salivate over companies with prestigious co-investors. “Who else is in the deal?” is either the first question, or the second, asked by investors. Sheep don’t like grazing too far from the flock. This way, if everything goes sideways, you can always point to others who made the same mistake and feel less stupid for losing tons of money. Looking to co-investors is also a lazy investor’s due diligence. Rather than doing actual diligence, which is hard, a lot of investors fall back on “if it’s good enough for other big-shots, it must be good.”

Startups also love touting big-shot investors they have. I can’t really blame them, but the data suggests co-investors have little-to-no predictive value when it comes to forecasting a business’s odds of life or death. Along these lines CB Insights recently reported that in 2014, 73% of tech exits were not VC-backed. If you insist on swimming with sexy co-investors you’re probably missing the boat.

It’s all about geography… nope.

Startups often define themselves by their location and there’s a sense that being in the right place (which is usually Silicon Valley) increases a business’s odds of stardom.

Here’s the reality, at least when it comes to venture-backed startups. On average, around 10% of venture-backed startups have exits regardless of where they’re located. That’s right. According to data from the National Venture Capital Association, around 9% of VC deals in Silicon Valley had exits over the past decade. Meanwhile, of the roughly 6,000 US-based startups that received VC funding between 2006 – 2011, 11% had exits. NY-metro startups had 12% exits. Los Angeles/Orange County startups had 8% exits. You get the idea.

What does this mean?

It means – proportionally speaking – success isn’t in the groundwater. Being physically located in Silicon Valley doesn’t somehow make your VC-backed exit any more likely. Your odds could be just as good, plus or minus a couple percentage points, anywhere else in the country.

If that’s the case, why does Silicon Valley have the lion’s share of venture capital-backed mega-hits?

Volume. That’s the answer. Silicon Valley may not have a higher percentage of wins, but it has the most numerical wins due to the sheer volume of VC-backed startups located there. Silicon Valley has roughly a third of the nation’s VC-backed startups, which is more than New England, Los Angeles, Washington, Oregon, Idaho, Texas, Colorado and the entire Southwest combined.

Not only are there more VC-backed startups in Silicon Valley per capita, but there’s a disproportionate number of venture capitalists there as well. Of the 25 most active US seed and early-stage VC funds, around two-thirds are based in Silicon Valley.

This is a bad thing for all but a microscopic cadre of entrepreneurs. VCs claw away to shower money on a tiny number of startups with the right teams (which, as you recall, doesn’t really matter) and the right co-investors (which doesn’t matter) that are located in Silicon Valley (which doesn’t matter either). For everyone else, it’s hard kibble.

Another unfortunate consequence of these myths – when combined – is bubble-producing behavior. For example, Silicon Valley pre-money valuations for seed and early-stage startups are typically 15% – 40% higher than in other US markets, even though the odds of a hit are roughly the same.

It’s all about a hot, huge market… nope.

My research has found an inverse (negative) correlation between the size and “heat” of a market and its long-term attractiveness for startups. For example, there are literally hundreds of startups trying to use big data to deliver more relevant, targeted ads to mobile users. While some have been successful, and others will likely continue to be, there’s a much higher rate of failure for such “hot,” highly saturated, intensely competitive markets when compared with smaller, less obvious markets with scant competition.

Remember when RFID was hot 15 years ago – droves of RFID startups chased Walmart as a customer and died crashing into each other. Meanwhile there were better results for RFID startups that started in odd, weird markets nobody considered sexy (ex. tagging sheep in pastures). If you’re still not convinced, consider Uber and AirBnB. Sure, they seem cool now, but 10 years ago taxis and budget hotels were a far cry from hotbeds of innovation.

I also haven’t found any predictive value in forward-looking financial forecasts. This won’t surprise most entrepreneurs, since it’s an open secret that forward-looking financials are basically creative prose in Excel format. They’re a nice fantasy, like daydreaming about how you’ll spend your PowerBall winnings. Let’s put it this way – I’ve never seen a dead company with a bad spreadsheet.

Then what’s it all about?

If you’re like other people I’ve shared this research with in the past, by now there’s a good chance you’re cursing us under your breath. We can all think of counter-examples and exceptions to the findings above. Yet I hope you consider how I arrived at these lessons – I’m simply describing the results of statistical analyses. I didn’t base these tidbits on personal experience or curated anecdotes. I’m not claiming to be a business guru. I’m just trying to observe the statistical nature of innovation as objectively as possible. This is the case even when – rather, especially when – data runs counter to conventional wisdom.

The outcomes I describe were taken from statistically significant sample sizes. That’s a key difference. Yes, we humans can feel the difference great leaders can make. It’s palpable.  However it’s another issue to form objective, falsifiable tests for what defines “good” or “bad” leaders and compare those to statistically meaningful samples of startups with objective, falsifiable definitions of “success” or “failure,” test for statistically significant correlations, form hypotheses of causality, test those hypotheses against random control groups, and so forth…  You have to count good leaders that failed, not just ones that succeeded. You’re responsible for the whole picture.

Doing this kind of thing, you often learn cherished truisms aren’t as universal or absolute as you thought. Sometimes they’re just dead wrong. When this happens, as a philosopher you nurse your ego. As an entrepreneur, you pivot. As an investor, you tweak your portfolio. As a data scientist, it’s just another Wednesday.

These are questions I’ve spent my life trying to answer, and will probably still be working on from my deathbed. I can’t profess perfection or spout absolutes. Yet what I have come to learn, at least statistically, is that the business model itself is the primary predictor of success or failure. Obviously, defining the “business model” is a mouthful, but for the sake of wrapping up this blog it’s the business itself that matters – not the team, not the investors, not the geography, not how “hot” the market is, not forward-looking financials.

It matters which customers you target; how you solve their problems; what competitors you encounter; what tradeoffs you break; how you operate on a day-to-day basis. Put simply, how do you make money? These are the factors – drab as they may be – that I’ve found the most predictive, statistically speaking. Like Warren Buffet says, “good jockeys will do well on good horses, but not on broken down nags.” Rid your mind of innovation myths. If all you do is pair a good business with a team that won’t screw it up, odds are you’re already focusing on better stuff than most.

This Post Has 6 Comments

  1. Jennifer

    We haven’t been able to raise money because none of us have big hits as founders and all our angel investors are smaller. How can we break through when VCs care about details that don’t matter. I agree they don’t matter, but we still need to raise money. A real hard spiral to get out of. Thanks!!

  2. Eric

    Hi Thomas–
    Yup. That’s why I like dashboards that incorporate
    both quantitative and qualitative metrics.
    You get a better balanced picture of the true impact
    of innovation as a “fluid” driver
    in a flexible business model strategy.
    Eric

  3. Kyle

    Good read, thanks.

  4. Mary

    The thing is, statistics can easily “prove” one thing or another, depending on the details. It’s all about generalisability, reliability and variability of the data.

  5. EW

    Hi Thomas,

    I am not sure if you read these comments but I just have a question on the study that I would appreciate your insights into.

    As a young VC with an actuarial background, I am still learning how to analyse and think about startups and so naturally, I find your work incredibly interesting. I just have a question around the point on teams. How did you define what a good team is? Off the top of my head, I would have thought that things such as education, gender, age, past employment history, size / strength of network, experience in the field would be the easy data points to collect and easy to run analysis on. But how did you look at softer elements, such as ambition, grit, likeability, passion, intelligence or is it assume they would be captured in the other more visible aspects (i.e. an intelligent person one can reasonably assume would go to a good university)? I am not super surprised by the conclusion that what is more predictive is just not having a bad team instead of picking a particularly good team but am interested in how this was looked at to help my understanding.

    Thanks!

    Eloise

Leave a Reply