In the data science culture, there are shared attitudes, beliefs, and assumptions; unless you're already part of that culture, you'll be surprised at how some data scientists behave. For instance, it might surprise you that there are bugs (or undocumented features) in the new software you just purchased, but the joke's on you — all the data scientists in the group expected bugs and think it's just a matter of finding them.
These behaviors are innocent — your data scientists aren't purposely trying to annoy you — and sometimes it's innocuous, but unexpected behavior is generally not something you want when you lead a team. To understand and subsequently align your data science team's behavior, you need to get them talking about their thought processes.
Getting data scientists to open up is harder than you might think. Most data scientists are introverts, so the vast majority of their dialogue happens internally. (I often have a full-blown argument with myself without uttering a sound — and sometimes I lose!) The good news is, in almost all cases, there's a very logical reason why they do things. The trick is to understand their reasoning. The only way to do that is to ask, though you may not get a clear answer.
The best technique for getting your data scientists to explain their reasoning clearly is to model the behavior. When you make a decision, meet with the team to explain your reasoning behind it. Not only will it bring your team more in alignment with your leadership, but it will also give them a basis from which to explain their reasoning. If you follow a consistent model for explaining your behavior, they'll pick up on the pattern, and probably start using it when you ask them to explain their behavior.
A good model to explain your decision-making process is the Issue, Rule, Analysis, Conclusion (IRAC) model; this is a common framework used by lawyers to build a case. Start with the issue, explain the rules that govern the decision, provide your analysis to support the rule, and then summarize with a conclusion.
For instance, if you decide to bring on a new analytic manager for the team, you might say, "Given my other leadership responsibilities, I'm concerned that I'm not providing enough management support for our team. To be effective as a team, we must be able to manage the complexities of all the moving parts, and ensure we have a good execution plan in place to move us confidently to our goal. I've found someone who has over 20 years in managing situations like this and she's got an excellent background in analytics. Therefore, it made a lot of sense to bring her on board to support our team. What are your thoughts?"
A time and a place for everything
To be successful with this technique, it's best to proactively set aside time for your team to talk to each other. If you have organizational change specialists in place, leverage their expertise to organize the agenda and facilitate the group. As a leader, I'm sure you've been to your share of these experiences, so just incorporate best practices from similar events that you've attended. At first, it may seem uncomfortable for the team to go through these touchy-feely sessions, but they'll come to appreciate the break.
It's important that you're very explicit about what you intend to accomplish with the team at these events: to share thoughts and feelings so you can come to a common understanding. Publish the agenda in advance and make sure they have time to process what their role is in the exercise. If you make this a frequent event (which I recommend), then you need less time for upfront preparation. After a few sessions, the team will develop better skills for articulating their thoughts inside and outside the facilitated team sessions.
Understanding the nodes in your data science team's neural networks (i.e., their inner thought processes) can be a challenge at first, but with a little work you and your team can get into the same groove. The key is getting them to open up about their decision-making process, which is not an easy feat with a team of introverts. Develop this skill within your team by modeling the behavior and organizing team sessions with the specific objective of sharing and communication. It's more for your benefit than theirs — they already know what they're thinking, and it probably makes a lot of sense. You're the one who is in the dark.
- What analytic managers can learn from neural networks
- Deep learning: A high-risk/high-reward big data investment
- Personality clashes stalling your data science team? Try the Myers-Briggs Type Indicator
John Weathington is President and CEO of Excellent Management Systems, Inc., a management consultancy that helps executives turn chaotic information into profitable wisdom.