Most executives I know operate on an exception basis. They set the plan and ground rules, and then only want to be notified if there's a problem. It's not a bad way to operate, and I've developed many executive-level processes that operate under this principle. But this approach can be dangerous with a data science team.
Silence = something's wrong
A data science team's volume can be an indication of its health. Data scientists aren't typically chatty, so a healthy team doesn't mean people are screaming up and down the halls, but there should be at least a murmur. When your data science team goes completely silent, something's seriously wrong.
You'll have a variety of personality types on your data science team, though it's safe to assume that your team's collective personality will be bias toward analytic. This is easy to validate by watching them.
My favorite personality assessment technique is Social Styles, which was presented by Robert and Dorothy Bolton in the mid 1980s. What I like most about Social Styles is that it's based on observed behavior and not a silly survey that attempts to dig into someone's brain. According to the Boltons, an analytic is unresponsive and unassertive, which is what you want in a data scientist — you don't want the core of your analytic competence drawing impulsive, capricious conclusions. There's no right or wrong personality, so there's no inherent downside to having a team with an analytic bias.
There is a problem when your team is stressed out — regardless of personality. Different personalities exhibit different behaviors when they're stressed, and analytics tend to clam up. It's dangerous when your team goes silent, because it's an indication that your team is stressed out. You better have a good team coach that can gauge the stress level of the team, and cool them down if the team is redlining.
Pump up the volume
You shouldn't be surprised to find out that some or all of the data science team members are under too much stress. There's a lot of pressure on most data scientists to perform, and many have an inferiority complex, especially young data professionals who are just entering the corporate landscape from college. A moderate amount of stress is good, but too much stress will shut down your productivity. It's hard to find the inflection point, though you'll have an indication it's out of control when your team struggles to produce results. It doesn't need to get this far though — silence is a good leading indicator.
If you notice less conversation and interaction within your team, immediately have a conversation with your team coach. If you do not have a team coach (shame on you), then enlist the services of an internal or an external expert on human behavior. They should have a background in small team dynamics, and they should be familiar with the way analytics function under pressure.
Your first objective is to validate that stress is causing the silence, and not something else. Your coach should ask team members (in confidence) how they feel about their job and if it's meeting their expectations. Your coach should determine if team members feel positive and engaged, or frustrated and inadequate. If you only have one or two people in the latter camp, then they may need some personal coaching; however, if your coach picks up on a pattern of unhappiness and despair, it's time for the team to cool off.
The best method for immediately bringing down the team's temperature is to pull them out of the fire. As much as you want to see the fruits of your strategy develop, pushing a team that's already redlining will only make things worse. Put the strategy on hold for a month or two while the team regroups. Data scientists love to learn new things, so send them to group training or bring someone in house to teach them something new. They also love solving problems without pressure, so let them drift in to pure research mode for a short while — this will get them working with each other, and more importantly talking with each other. When the sound level increases and you see smiling faces again, it's time to get back to work.
It's easy to prevent your team from an overstressed situation by watching for the best leading indicator: silence. When it comes to data scientists and other analytic-minded people, no news is usually bad news. A little alone time is quite normal, but when everyone goes dark for an extended period of time, it's time to get worried. Rely on your team coach for effective diagnostics and interventions, and talk to your analytic manager about letting off the gas a bit.
If your corporate strategy relies on your data scientists, and you don't get them talking again, your whole company may go silent.
- Identify morale problems in your data science team with the help of sentiment analysis
- Screen data artist candidates for potential creatives vs. pros personality clashes
John Weathington is President and CEO of Excellent Management Systems, Inc., a management consultancy that helps executives turn chaotic information into profitable wisdom.