Top Innovator John Thompson: Why Managing Analytics Teams & Data Scientists Is a Unique Problem

Has the management of data science teams kept up with the meteoric rise of the discipline? A recent Domino Data Lab ebook, The Data Science Innovator’s Playbook, explores new approaches and thinking on the problem, from top data science innovators who have had a hand in building some of the world’s most successful teams. Some of them say that managing data science and analytics is a distinct, new problem that demands fresh approaches and rethinking old ideas learned in managing other types of teams, such as IT, operations personnel, and software developers.

“Data science teams are unique,” says John K. Thompson, author of Building Analytics Teams and a leader and builder of several highly successful enterprise data science teams. He was recently featured in the Data Science Innovator’s Playbook for his work.

“They’re not development teams; data science teams are a creative function, and have very little in common with software development projects and related work,” he says. Rather, he says data science teams have an innovation function and a strategic function, not an operational one.

The Optimized Data Science Team

That’s why, he says, effective data science teams should be managed with a degree of freedom to experiment and fail. Plus, they’re optimized when they have a mix of young, old, and curious staff members.

The approach Thompson says works best is to understand what type of team you need now, for your business. In his book, Thompson outlines three different types of teams:

• Artisan - Each data scientist is responsible for all aspects of their projects from design to presenting to senior executives
• Factory - All work flows through specialized sub teams. Those teams work sequentially to execute specific tasks in the overall data science process
• Hybrid – You have both Artisan and Hybrid teams working together

Managing Analytics and Data Science Teams Changes Over Time

Also important to realize, Thompson says, is that the best organization for data science and analytics teams usually changes over time.

“Regarding my current team, I started with an Artisan team,” he says. “ I added a Factory team a year ago. Currently, I have a Hybrid team. Data science teams are dynamic, so the team organization should be dynamic as well.

And addressing the positive developments in the field, Thompson says both businesses and data scientists benefit from the current state of data science and analytics, including the emergence of data science tools such as enterprise MLOps.

“We have reached a market state where we have solid foundational tools and technologies to deliver robust models and applications,” says Thompson. “We have a shortage of analytical talent, but we have enough talented data scientists to drive change and impact across companies and industries. And, we have open access to significant data assets to use in our efforts.”

Download the free The Data Science Innovator’s Playbook to read more insights from Thompson–as well from as many others–on themes, strategies, tactics and insights for managing teams and developing innovative analytics programs. This exclusive content includes interviews from:

• Cassie Kozyrkov—Chief Decision Scientist, Google
• Andy Nicholls—Senior Director, Head of Statistical Data Sciences, GSK plc
• Mona G. Flores—Global Head of Medical AI at NVIDIA
• Robert Nishihara—Co-creator of Ray, and Co-founder & CEO, Anyscale
• John K. Thompson—Analytics Thought Leader, Best-selling Author, Innovator in Data & Analytics
• Glenn Hofmann—Chief Analytics Officer, New York Life Insurance Co.