The Man Who Knows Whether Any
Startup Will Live or Die

BY KLINT FINLEY
Ben Miners/Getty Images
Starting a business is a dangerous thing.
A larger competitor might undercut your prices. Someone might sue you for patent
infringement. Someone else could sue you because your products don’t do what you said it
would. Or, well, the market may have no interest in what you’re selling. According to the
U.S. Bureau of Labor Statistics, about half of all businesses fail within five years.
But Thomas Thurston thinks data science could remove a fair amount of the risk. For the past
nine years, he’s been honing techniques for evaluating business plans statistically rather than
intuitively. He calls it business model simulation, and you can think of it as something akin
to Moneyball for investors.
He says his simulations correctly predicted that Snapchat, Uber, and Airbnb would be big—
and that they’re now right about 66 percent of the time when predicting that a company will
still exist within five years. When predicating that a company will fail, he adds, they’re right
88 percent of the time.
The simulations have proven so successful, Thurston is now using them to make money for
himself. He runs a research firm called Growth Science, which sells his predictions to large
companies—and applies them to investments he makes as a partner at the venture found
Ironstone Group. In the long run, he believes, these simulations could have a rather profound
effect on the business world as a whole—because they can steer people away from bad ideas.
“Most businesses fail, and that’s not good for people,” he says. “People lose their jobs, the
economy suffers.”
Thomas Thurston.
Growth Science.
He admits the models will never be perfect, but thinks that even a model that’s only right
about 50 percent of the time could help investors and entrepreneurs avoid particularly bad
ideas that, to the untrained eye, look like excellent opportunities. If fewer businesses fail, he
reasons, the whole economy would be more stable and everyone would benefit.
Thurston isn’t alone in applyingMoneyball-style data science to investing. Google Ventures
takes a data-driven approach, as do funds like Correlation Ventures and Venture Science. But
he isn’t just using his calculations to make his own bets in the market. Growth Science also
helps big corporations on investments, acquisitions and strategy. 3M, for example, uses it to
predict the degree of success of new product and services. The idea is to help these companies
make informed decisions and avoid having to do mass layoffs. And eventually Thurston
thinks it could help small businesses and startups as well.
Banishing Intuition
Thurston came up with the idea to simulate
business models 2006 while working for Intel
Capital, the investment arm of the venerable chip
maker. One day he decided to chart Intel’s
investment history and see if any patterns
emerged.
His approach is based on turning various pieces
of qualitative information—such as whether a
company is a “first mover” or “fast follower” in
a market—into quantitative data that he can plug
into a spreadsheet. That requires a degree of
human judgement, but this also requires a certain
amount of rigor or consistency.
“You can’t trust the model until you get all the
intuition out of it,” Thurston says. “The hard part
of that is translating the qualification into yes or
no questions,” he says. “How do you define the
market? How do you define first mover?”
Surprise, Surprise
Using this process, he discovered some
surprising things—most notably that a
company’s team is only about 12 percent
Thomas Thurston’s Top Three Bets
Arcimoto: An electric car company aiming to
offer a two-seat vehicle that can travel 130
miles on a single charge at a much lower price
than competitors like Tesla and Lift Motors.
“Arcimoto is targeting the lowest cost
automobile platform to own and operate in the
US, with the simplest possible solution,”
Thurston says.
Color Genomics: A startup building a system
designed to help the masses take advantage of
genomics. “Color is using computing and data
to do this in a much more simple way at a tiny
fraction of the cost that’s accessible to anyone,”
Thurston says.
Indow Windows: Swapping out your drafty old
windows for new energy efficient ones could
save you a bundle in the long-term, but not
everyone wants to spend the time and money
to retrofit their entire home or office building.
Indow Windows offers inserts that can improve
efficiency without the cost or hassle of
replacing the windows entirely. “Some other
startups have tried this, and some of the big
guys are trying to respond, but there’s a lot
more innovation required to pull this off than
most people suspect,” Thurston says. “In a very
short timeframe Indow has zoomed up to
become the market leader.”
predictive of a company’s success. “You need to find a good team that won’t ruin the
company, but hiring ‘rock stars’ isn’t that great,” he explains. The market the company is
entering is far more important than who’s running the company.
His work at Intel ended up landed him a Harvard University fellowship thanks to Clayton
Christensen, author of influential book The Innovator’s Dilemma. After the fellowship, he
started Growth Science to fund the further refinement of the process, and bring it to the rest of
the world.
To The Masses
Thurston wants Growth Science to advise entrepreneurs—and help people with good ideas
find better business models. And although his work has mostly been used by large companies
and investors so far, he says, it’s beginning to trickle down to the entrepreneurs themselves.
Last year, for instance, Ironstone Group invested in electric car company Arcimoto, but the
company barely made the cut. “We liked them, but they were on the edge,” explains Thurston.
So he tweaked his simulation, and eventually decided the company should go after emerging
markets rather than just the U.S. For Arcimoto founder Mark Frohnmayer, that was a crucial
piece of advice.
“We have had a strong interest in emerging markets from the beginning, because this is a
global problem that we’re trying to solve,” Frohnmayer says. “But we’ve been doubling down
on the emerging market story in the past year, making sure that we had an offering that would
be competitive not just locally but in the world market.”
The Problem
Even businesses that Thurston has ultimately turned down for investment purposes, he says,
have ended up benefiting. “People will come back to us months later, and say: ‘We thought
about what you said, and now we’re doing something different.'”
But what Thurston would really like to do is help all businesses, not just the ones Ironstone
considers investing in. The problem, however, is that Growth Science charges a few thousand
dollars to consult these companies because it still take a lot of time to convert a traditional
business plan into something the Growth Science team can run through their algorithms.
That’s still too much for most early stage companies to spend.
One way to make it affordable would be to automate more of the process, and offer it as a
web-based service for a low monthly fee—or maybe even for free. And, in fact, Growth
Science has already built a beta service that does just that. But there’s a catch.
According to Thurston’s own model, Growth Science’s own chance of survival following its
current business model is about 69 percent. Adding the automated service would actually
improve its chances, he says. But that would mean risking cannibalizing the already
successful business he’s built consulting higher-end clients. In short, he has an innovator’s
dilemma of his own. And that goes to show that there’s always risk in change, no matter how
reassuring your data models are.