Dan Zarrella of HubSpot, Research Interview on Data Science

I had a fascinating talk interview for the What is a Data Scientist? series today with Dan Zarrella, Social Media Scientist at HubSpot. Zarrella recommended that data scientists should be prescriptive in their research. Instead of doing research for its own sake, it's better to start with some sort of question about how you would want to change behavior, for example: How to get more twitter followers? Then find correlations until you have some good hints about causation.

Zarrella also loves mythbusting. It turns out that in social media marketing the idea of engaging in the conversation doesn't really get you more followers. Getting people to comment on your blog doesn't get more views.

In his book, "Zarrella's Hierarchy of Contagiousness", Zarrella makes great use of a framework for how ideas propagate that has three stages: Exposure, Attention, Motivation. He was inspired by other frameworks from marketing and the military. The power of a framework is that it helps organize thinking around a specific goal. It is easier to avoid research for its own sake when you are working in the context of a goal-oriented framework.

Zarrella also recommends that a good data scientist should not be afraid of being wrong. In academia, being wrong is a death sentence. Not so in the world of business where the clockspeed of analysis is much faster and the recommendations are tested through action. Shooting for a good batting average as a data scientist is better than attempting to be being perfect.

Zarrella also had some good thoughts about how technology leaders can overcome the limitations of the Raw Technology Persona. He recommended preparing to speak well in public, even if you hate it, as a key leadership skill.

I'm looking forward to writing two or three stories out of this interview. The first will be part of the "What is a Data Scientist?" series on Forbes.com.