March 17th, 2014

Will data eat VC? ( or, why we hired a data scientist )

"A data scientist is a statistician who lives in San Francisco. Data science is statistics done on a Mac" - @smc90

This week we welcomed a new member to the Balderton team, Ferenc Huszar, ( @fhuszar ). Ferenc has a great background to work in VC, with experience in the VC backed start-up, PeerIndex, and impressive academic credentials ( for those interested in Adaptive Bayesian quantum tomography ). However Ferenc joins us as a data scientist, a new role here. Some VCs already have data scientists in their team ( Hilary Mason and DJ Patil to name a few ) and it looks like there will soon be more ( General Catalyst job listing here ). Capturing, understanding and using data is beginning to change venture funding as it has so many industries. But does this mean we're all going to be out of a job?

Is data going to eat us?

Balderton, like many funds, has always used tools to better understand markets and identify interesting companies. Since I joined last year, we've built a set of proprietary systems that automate our use of private and public data. In fact there has been an explosion of these tools performing similar tasks such as Mattermark and Inkwire.io in the US, and Bright*Sun in Europe. These tools track Alexa rankings, twitter followers and similar metrics to identify 'fast-moving' companies and even suggest new investments based on your portfolio history. These tools aren't changing the way we work yet, but they are making some of the basic work we do much quicker. Effectively, they're supercharging VCs. But is this the beginning of the end for the human element? Will these tools replace us and is this good for the industry?

People don't scale

What these tools and traditional methods can't offer, and what Balderton focus on, is the insights gathered from backing 200+ founders over 13 years of investing. Unlike the companies we invest in, venture funds aren't scalable, because venture is a people based industry, and people don't scale. While there are more opportunities now to invest in breakthrough technology companies than ever before, the best VCs will still limit themselves to supporting a set number of companies, as their time and energy is finite and to take on more would dilute the value they can offer.

The threat of 'Momentum Capital'

What we are not doing is building an algorithmic approach to investing, following the hedge funds and equity traders of the financial world. The risk in the growing use of tracking tools like those mentioned above is that investors start relying on short-term signals and chase momentum, skewing what founders do to get their attention.

A bump in your AngelList signal or Mattermark score may be nice, but it shouldn't be something to optimise for. Yet if enough capital starts following these metrics then founders may feel compelled to follow suite, and this would be a nightmare for the tech industry. CEOs of public companies complain that pressure from equity traders forces them to focus on shorter term outcomes over harder, strategic decisions ( the average holding time for a trade on the FTSE is <9 days ). The beauty of venture investments is they are supposed to avoid this, as partnerships of many years that last through the ups and downs. Most investors arn't interested in backing the app that got the most twitter followers this week, but the founders with the biggest vision. Ferenc's work at PeerIndex and elsewhere on understanding the dynamics of people and networks, and filtering out the trivial from the substantive, chimes with what we think is the key to successful venture investing.

And?

Then why hire a data scientist? What are they going to do?

The type of question we want to answer isn't 'Who got the most downloads on the App store this week ', which is what a lot of these tools track, but broader questions such as ' does it make sense to focus resource on iOS at all? How can we do it most successfully? Who can help our companies do it?'. These questions require a deeper understanding of both data and people.

We are already sharing the outcome of this work with our portfolio, and our goal is to strengthen the tools and insights we provide to help them make better decisions using the abundance of public data available. We also want to share these learnings with the tech community, and be part of broader conversations on product development, hiring teams, and cracking conversion. As part of this effort Ferenc and the Balderton team will be publishing regular pieces on different topics and running workshops across over the next year. We'd like to hear from you about areas we should cover.