Since the early days of venture capital, the investment process has been based on a combination of gut, people and deal timing. And while data has always been a part of that equation, it never took center stage.
Recently, however, we’ve seen the emergence of quantitative venture capital. Funds and angels collect data on startups and use that data to guide deal scouting and the decision making process. At Disruption Corporation, we stand on the thesis that data can help investors make better decisions. Over the past year, we’ve been working on tools to support that thesis - one of those tools being Indicate.
During this process, we’ve had some time to think about some of the problems when dealing with data and making data-informed decisions. Here are some of the things we learned:
- Be wary of confirmation bias: humans are hard-wired to see patterns everywhere. And we do - often where they don’t exist. Keep in mind that any person (and any system) may represent data in a way that makes it hard to see a problem.
- The right metrics for the right company: no two data points are alike. If a mobile app company sees a ton of traffic to their website but app downloads don’t follow a similar trend, that might mean a lack of market fit. Make sure you’re paying attention to the right numbers.
- Public data is limited: pre-investment data sets are often based on public data (Alexa traffic estimates, social media attention numbers, App Store downloads). Consequently, important metrics such as revenue are often omitted. Always evaluate the data you have, but do not ignore the unknown unknowns.
- Gaming: if there’s a financial incentive behind numbers, some people are going to inflate those numbers. Traffic, social media numbers, and app downloads can all be gamed. The same way data supports the investor’s gut, the same gut should confirm data.
- It’s a slow process: data-backed decisions need a validation feedback loop to confirm hypotheses. You make a decision based on a set of metrics, wait for the outcome (e.g. an exit), and validate the initial model. Venture capital moves at a slow pace, so this feedback loop is slow too and there isn’t a lot of history to work with.
- Black swans: there are always going to be surprises (both positive and negative) that data can’t predict. And as Nassim Taleb describes in The Black Swan, these surprises are often rationalized in hindsight. You should be okay dealing with the unexpected (and frankly, that’s one of the requisites of being in venture capital).
These are still the early days of quantitative venture capital, and as an industry we have a lot to learn. Over the next few months, I’ll be posting more about how we use data at Disruption Corporation, and how you can use it too, whether you are investing in startups, or running one.
I’d love to hear your thoughts. Feel free to reach out on Twitter.