BookmarkSubscribeRSS Feed

Analytics Leadership from a data scientist perspective

Started ‎04-03-2023 by
Modified ‎09-19-2023 by
Views 828

With a background in Civil Engineering and Econometrics, Paul Koot was team captain of Notilyze’s winning Hackathon team in 2020 and 2022. In his role as data scientist he has gained extensive knowledge of machine learning models in manufacturing projects and text analytics. In this article we ask him about analytics leadership and what this means from his experience.


1. How would you define analytics leadership?

Recently I encountered the following quote about leadership from Dwight D. Eisenhower: "The essence of leadership is to get others to do something because they think you want it done and because they know it is worthwhile doing." I would say the following for analytics leadership: The essence of analytics leadership is to get data to do something because the customer thinks data can do this and because you know it is worthwhile doing. What I mean by this is the following.



A first condition for a data analytics project to succeed is that a customer should be convinced that data can help with this problem. Most of the time, this is (at least partly) the case, as the customer would not have approached you (the analytics leader) in the first place. However, it is not apparent that all stakeholders in your project are aligned on what value data can bring to the table in solving business problems. As an analytics leader, a very important part of the process is to make sure all stakeholders acknowledge the business problem and acknowledge data can help solve this problem.



Then, the second part is critical. You, as an analytics leader, should realize that it is worthwhile spending effort and time in using data to solve this problem. And more importantly, you should find a way data can tackle this problem efficiently. You, as an expert, should clearly separate the best solution from the solution the customer thinks they need while keeping track of the customer’s business knowledge.


This business knowledge is essential for a solution that will not stop after finishing the Proof of Concept phase and really adds value after implementation. In tackling the problem efficiently, it is important to realize that data analytics must always be the means to an end, not the goal itself. Realizing this helps to separate the customer’s problem from the technology hypes they think they should also start using. This means you can sometimes solve the problem more fundamentally than the customer initially thought (you can read more on how to do that in this article). However, reformulating the business problem might lead to another iteration of aligning all stakeholders to acknowledge this. In my opinion, an analytics leader is capable of doing so.


2. What different roles do you see in a team of data scientists?

I see three essential roles at the core of any project. The first is the business translator, who has the skills to understand the business problem and explain to the customer why that is the business problem. The latter is important; if a customer does not agree with your problem definition, they will never agree with your solution. Also, the business translator splits the problem up into different more minor problems. Each problem can be tackled separately, leading to actionable insight.


The second role is the data engineer; this person assembles, transforms and cleans the data. The ETL process (Extract, Load, Transform) is often not visible in the final product, but it is at the core of almost any solution. Data engineers deliver the Analytical Base Tables (ABTs).


The third role is the data scientist, who translates the data into actionable insights. Often multiple iterations exist between a data engineer and data scientist, as decisions on how to show results on a dashboard affect the most efficient way to store things in the ABTs. An analytics leader could fulfill any of these roles, but naturally, the role of business translator comes with most of the customer interactions and therefore would be the most obvious role for an analytics leader. But note that one person can have multiple of these three roles, depending on the project size.


On the other hand, there could be more roles than just these three. To meet the customer’s wishes, there could be software developers who deal with integrating the solution in the existing architecture or creating a new tailor-made application. UI-designers ensure the interface is user-friendly and point out essential insights. And cloud engineers ensure this all happens within the cloud in a scalable and secure way.


3. What is your experience how a Data Scientist should work best with the stakeholders?

At the start of a project, the full focus should be on the business problem; a data scientist should try to understand what is at stake (this often is more than ‘the problem’), who has to give a go/no-go, who experiences the problem, who needs a solution (not always the same person!), who will work with your solution. Every hour of this orientation will save multiple hours in the following steps. After this first assessment, the data scientist must reflect critically; do you agree with the initial problem statement, or should we focus on other things first? If the latter is the case, the next step is to convince the stakeholders the focus should be on something else. Point out what you see as the problem. Only when everyone agrees on this, you can move forward with proposing your solution.


4. What are the required key skills a data scientist should have to be successful in the job?

That depends on the setting. A solo data scientist solving a problem needs different skills than a data scientist who is part of a dedicated scrum team with all kinds of roles. But skills that are always useful:

1) be critical of yourselves on why you are doing things, as you might be the first one asking that question;

2) keep in mind the operation phase when developing a first version;

3) math/statistics OR domain knowledge: you need to be able to add a critical side note either from the business case perspective or the data/algorithmic perspective. Both of these knowledge fields need to be in a project, so if you don’t own them, make sure you find someone else as your counterpart.


Well, well: this is a fun, informative article on data scientist leadership.  Thank you for sharing your words of wisdom, Paul! 

Thank you Paul, for your great insights into the different types & roles of leadership in the Data Science environment

Version history
Last update:
‎09-19-2023 03:55 AM
Updated by:

Ready to join fellow brilliant minds for the SAS Hackathon?

Build your skills. Make connections. Enjoy creative freedom. Maybe change the world. Registration is now open through August 30th. Visit the SAS Hackathon homepage.

Register today!

Free course: Data Literacy Essentials

Data Literacy is for all, even absolute beginners. Jump on board with this free e-learning  and boost your career prospects.

Get Started

Article Tags