Managing data teams for the first time: 6 things you should know

3 min readJun 29, 2020

Scale your impact but keep time for learning and doing

Some things I learned managing data in tech. Common sense but don’t underestimate them.

It is not really about managing

Managing is just about directing the right resources in the right place. Really in Data, it won’t mean a lot. For instance, knowing on which project to work will be a lot more important as well as showing leadership. Data organization is not an operation department. Organizing tasks in a scrum/agile way is important but it should feel natural to manage this rather than be draining on your time.

Managing in data is about setting goals, incentives, and the right structure. When you have such dedicated and high talent.

It is not because they only want tech mentoring that they only need tech mentoring

Technical excellence is everything in tech and it can sometimes feel that improving tech skills is the only way to improve results. Well, it is the case for junior profiles, but for senior profiles, part of the scalability comes from the way they think of a problem and the strategy they put in place to solve them. It also comes from the ability to find the right problems. Yes… so many companies hire data folks to find data problems. Everyone is hiring them, we must hire a lot of them too. Data is such a cutting edge domain that it leads to having to look for the right problem and the skills needed for that are close business understanding and creativity.

Don’t get amazed by fancy solutions from the team

Some fancy formula… don’t fall in love with them

Your team member will come back to you and say… “I just used reinforcement learning and was able to increase marketing performance by 1.54%” Sorry, what was the problem you tried to solve? O… we just needed a heuristic to realize in 2 weeks without complex engineering. Well, this typically does not reach the goal.

This is actually not so easy once you manage teams because they get into fancy and technical things but you still need to deliver as a company and as a team. Balance deliverable and intellectual stimulation.

Set a direction not a solution

Basically, this one is obvious and is just about avoiding micromanagement. The whole process of solving a data problem is the exciting part of data science and analytics. Not prescribing a solution does not mean you are not part of the solution nor that you should not influence it by other means. For instance timeline or and prescribe milestones can avoid some solutions that would be unwise for the company.

Find the right level of involvement in coding

You must spend some time getting hands-on and understand what is happening. The Tech sector is quite particular in this respect, you will find very few managers that just manage people. A manager is rarely a job in tech but more a role in addition to IC. This is especially true for first-level managers.

Do get involved in execution. You do not have to drive every project but do mentor on projects and get involved on an IC level for some of them. Your team needs to breathe and needs empowerment.

Find the right level of involvement with stakeholders

Stakeholder engagement is important when managing a team. You should streamline communication and progress on a regular basis. Position yourself as a thought partner rather than an execution partner.

What is the difference besides looking like a though leader? (I don’t like this title)

Execution partner means that your business counterpart asks you to solve problems that he thinks is the right one.

Thought partner means that your business counterpart has an ongoing discussion with you to identify problems at a strategic level and you are the one coming up with a solution.

No need to say that though partner can identify deeper problems and have more impact.




Strategy/Data/Leadership head of DS at OCBC ~~ exTwitter ~~ ex-gojek