Is it possible to predict the next unicorn company through data?

Google+ Pinterest LinkedIn Tumblr +

Predictive analytics is one of the most successful applications of AI today. It’s used in almost every field, from music (Spotify’s Discover Playlist), to policing, to medicine (Google developed a tool that can predict cardiovascular risk factors). Algorithms are categorically superior than humans at identifying patterns and applying them to future scenarios based on thousands, if not millions, of data points. It seems logical, then, that an algorithm should be able to do the work of Venture Capitalists — after all, they’re just making a highly informed prediction.

According to a Partner at a Silicon Valley mid-sized venture capital firm, 85 percent of predictive value comes from quantifiable factors including the market, competitors, trends, timing, and addressable customer potential. “VC’s love to talk about how they invest in people, not companies,” the Partner explained. “But the reality is that only about 15 percent of the equation has to do with the team and founders. That’s significant, but it’s nowhere close to the whole picture.”

Conceivably, then, predictive analytics could quantify 85 percent of any given investment decision — and it could give more than just a simple “invest/don’t invest” decision: it could predict 10-15 year valuation, venture rate of return at different investment figures, and make predictions about the potential market share a company could get based on competitor figures.

Some companies are already leveraging the power of predictive — but is it working?

In 2015, an early-stage VC firm called SignalFire launched in San Francisco. The firm, which has Google and Yahoo Machine Learning experts on its team, uses analytics that track millions of data sources in real time, such as monitoring capital flows into startups and the movements of key employees. It’s not the only one doing this: Correlation Ventures, for instance, uses proprietary predictive analytics for decision making on investment deals — which it claims cuts time-to-decision by weeks. NFX Guild, an investment firm that also has an accelerator, identifies new investors for its startups through a proprietary software. Google Ventures was using analytics to make investment decisions as early as 2011.

But is predictive paying off?

A NewtonX expert, previously a successful venture capitalist and now a finance and Silicon Valley journalist, explained how to understand any conclusions drawn about predictive analytics in VC investing:

VC funds typically take 10-15 years to return money to investors. To achieve a venture rate of return (how a “good” investment is defined), a fund needs a 3x return on a ten year investment — so a $100M fund would need $300M return. Only a very small percentage (roughly 5 percent) of funds achieve this. The average gross return for all dollars invested into companies exiting in 2016 was 2.2x, while 10% of these dollars realized a return of 4.9x or higher, according to an index done by Correlation Ventures. That only 10% realized a high return doesn’t mean that predictive analytics aren’t working — it means that it’s too early to tell. “We only have a maximum of ten years since predictive analytics started being used in investing. That’s not enough time to draw any conclusions,” the journalist declared.

VC funds make the most off of a small percentage of companies. Judging by the math above, VC investing isn’t highly profitable unless you have a few portfolio companies that make it big, like an Uber or Airbnb. Let’s look at whether the VC funds that rely heavily on predictive analytics are successfully identifying unicorns:

Google Ventures has invested in almost every unicorn you can think of; a few of the most successful include Uber, Hubspot (IPO’d), Jet (acquired), Nest (acquired), Slack, and Stripe. It invested in Uber in 2013 (four years after it launched), and its initial investment gained 14 times its value over the next three years.

Correlation Ventures was founded in 2012, and while it’s had 23 exits, it hasn’t had any unicorns. That said, the company has publicly stated that it values “triple crowns” (companies that realized a cash-on-cash multiple of at least 10x, realized an IRR of at least 100%, and had VC investment of at least $1M) over unicorns — a bold statement considering that triple crowns are even more rare than unicorns. Correlation has invested in companies including Casper Sleep, Sun Basket, Personal Capital, Virsto (acquired by VMware), Optimizely, and Lever.

NFX Guild has invested in a few unicorns including Lyft and DoorDash, as well as several other highly successful companies, including Trulia (acquired), Patreon, and Poshmark.

Finally, SignalFire has not invested in any unicorns — but considering how recently it was founded (2015) it’s too early to judge.

Time is of the essence — the real value of predictive is speed

Identifying the next unicorn company is only part of the equation for VCs, according to the NewtonX expert who is currently a partner at a Silicon Valley firm. The other part is landing the unicorn.

A few days can mean the difference between winning and losing the deal. The firms that identify the next unicorn fastest will get the chance to invest in them. Machine learning can also dramatically reduce the time that VCs spend on due diligence: it can analyze algorithms for efficacy and novelty, and rapidly compare numbers (MAUs, session length, etc.) to the numbers at the same lifecycle of apps that ended becoming unicorns. This information can lead to a pivot in positioning, valuation, or strategy. And while every VC already does this, analytics can do it a lot faster than humans can.

Chris Farmer, Founder and CEO of SignalFire chatted with NewtonX for this article, and elucidated where he sees the value in analytics: “We’re a hybrid system that still includes venture capitalists to make the final decisions and balance quantitative inputs (e.g. performance metrics) with qualitative ones (the vision or grit and determination of the founder/team). Great systems are not enough; you need top quality investors and experts as well, and the combination of the two is optimal – augmented intelligence vs AI.”

Will predictive be able to identify the next unicorn more accurately than humans? It’s unlikely. But it may be able to identify it faster, and perform due diligence more completely and rapidly than humans can. While even the most data-driven VCs still need humans to make the final decisions, the data-driven predictions that analytics can offer make these choices smarter and faster. And for VCs, this could be the make or break factor for success.  


About Author

Comments are closed.