Building a successful data science team

Table of Contents

Sign up for our newsletter

We care about the protection of your data. Read our Privacy Policy.

Everyone wants to have cohesive teams that achieve results – but how can it be done with data scientists who each work on their own projects and interact in different ways?

There are three factors – business, technology, and mathematics – that are the trifecta of effective data science. Business gives you the problem, technology and mathematics bring the techniques that make sense of the data. Each of these factors need to be considered when framing a data science team.

This data science life cycle portrays the steps that data must go through time and time again until the customer accepts it.

On top of this life cycle, there are pillars of a good team. We have customized these pillars for a data science team by drawing inspiration from the best-selling book The Five Dysfunctions of a Team by Patrick Lencioni. Trust forms the foundation of any successful team. Trust is earned on an individual basis and requires the team’s effort to achieve. Once a team has cultivated trust, healthy and productive conflict can exist within the team. Through conflict a team is capable of accepting each other’s personal vulnerability that in turn brings problem clarity that would not exist otherwise. Only after clarity is reached, your team can focus on build models and solutions to address the business needs. Quite often your team will still find that their initial solutions fall short of business accountability, leading to a healthy cycle of iteration and innovation. Finally, the process will galvanize your team when their results are effectively shared with stakeholders and leaders who will recognize the teamwork that provided business success through data science.

What does it look like to have trust within a team?

  • Being unguarded and genuine with one another
  • Giving one another the benefit of the doubt rather than jumping to conclusions
  • Asking one another for help and input regarding your areas of responsibility
  • Apologizing and being open about weaknesses and mistakes

With data science and teams, there is an interplay between trust and healthy conflict. Debate around that conflict is crucial. The conflict cannot take place without the trust. In our team, there are different personality types that are learning to work together. As we learn more about each other, we are able to trust each other more.


How can you have productive conflict on a team?

  • Voicing your opinions even at the risk of causing disagreement
  • Seeking out your teammates opinions during meetings
  • confronting and dealing with the most important – and difficult – issues
  • Exploring everyone’s ideas to uncover the best solutions

The personality spectrum is where you need to find out what your type is and where each team member falls on the spectrum. There is one end with the spirited debater, who tends to be expressive and shows convictions during debate, and on the other end, the calm debater, who tends to use measured dialogue and remains largely unruffled. After that has been done, one can figure out how to work together and develop that trust.

Clarity is also important. You need to resolve conflicts internally and decide that as a team, we are accomplishing this specific goal. Coming to one decision means that not everybody gets their way – but because of that trust and conflict, each person is heard.    

What does it look like to have clarity on a team?

  • Supporting group decisions even if you initially disagree and embracing some risk
  • Being clear about the team’s overall direction and priorities
  • Ending discussions with clear and specific resolutions and calls to action
  • Leaving meetings confident that everyone is committed to the decisions that were agreed upon

Overall, as you get a team moving, people get more comfortable and their voices are heard. If you want to have meetings that create results, rather than having meeting to have meetings, there needs to be accountability. As a member of the team, you are accountable for a piece of the work.

If you are doing work simply because your boss said to do something that you do not agree with, your commitment suffers and the final product suffers. All of these pillars lead to team members being bought into the idea of what you are doing as a company. Feeling pressure from your peers is a nice way of working because you do not want to let the team down. You then feel good about accomplishing your part.


What does it take to have accountability in a team?

  • Finding areas of actionability in the solution being deployed
  • Feeling pressure from your peers and the expectation to perform
  • Confronting peers about problems in their respective areas of responsibility
  • Questioning one another about current approaches and methods

The culmination happens when you get to this point – the business results. Everything comes together and the results are shared with the business stakeholders. New ideas need to come out of a product that is finished. The scaling needs to happen vertically and horizontally. The evangelization of the work that has been done is of critical importance at this point. When looking inwardly, how is the team responding to it? How comfortable is a team member at the end of the project? They are typically different at the end of a project. There are team members that take feedback well and tend to get better the next time around and will make improvements. As team members learn more, they are more comfortable taking risks and they start in a better place for each project.

How can you share your business results?

  • Valuing the business solution and its impact versus the technology stack being used
  • Willingly making sacrifices in your area for the good of the team
  • When the team fails to achieve collective goals, taking personal responsibility to improve the team’s performance
  • Being quick to point out the contributions and achievements of others

    In a sample individual profile of a data scientist, there are various ways that an individual may be. This is where we begin to see the results of what we have done up to this point. We start looking at the numbers and metrics. We need to look at the key profile. It all starts at the base – and the agile methodologies will help with the day to day. Use these pillars to measure where you are now and where you want to be in the future. A high-level vision will help you get more data driven as an organization. We have now evolved from simply being individuals to a team as a result of establishing each of these pillars.

Keep reading

Dig deeper into data development by browsing our blogs…
ProCogia would love to help you tackle the problems highlighted above. Let’s have a conversation! Fill in the form below or click here to schedule a meeting.