Which departments work with data scientist teams

How to successfully build a data science team

I was recently allowed to write a guest article for the freelancer platform paints on the topic of how to build an outstanding data science team in a company. In this article I summarize the essential points.

Why should companies build their own data science teams?

A uniform trend has been evident around the world for several years. The companies with the highest turnover have an above-average level of competence in handling data. These companies include, for example, Google, Apple and Tesla. These companies use data to make strategic decisions and have managed to develop lucrative data business models. These companies were able to develop an extraordinary competitive advantage in this way. For such a strategic orientation, data competence must be broadly anchored in the company.

Of course, companies can also buy data services. In the long term, however, it makes sense for many companies to develop a certain degree of independence from external service providers and to equip their own employees with the valuable skills. Because this not only leads to new innovations more quickly, but also promotes a data-driven mindset across the various departments.

Since the implementation of this strategy is more like a marathon than a sprint, most companies start by setting up their own data science team.

What roles does a successful data science team need?

There is no general answer to this question. What is more important is which tasks the data science team is supposed to perform and how much preparatory work has been done, especially with regard to the provision of data.

Many companies start with the recruitment of a data scientist as they take on the primary role of data modeling. However, very often data scientists initially hardly spend any time modeling data because the data is either insufficiently available or of insufficient quality. In order to handle these tasks, a data engineer should be included in the team right from the start. Its task is to combine, cleanse and standardize data from different sources.

In addition, a data analyst should be part of the data science team from the start. A data analyst is usually a little less involved in data modeling, but usually has a little more business understanding and communication skills. Data visualization in particular is one of the tasks of the data analyst. This enables the data science team to effectively communicate its results to stakeholders.

To ensure that the data science project produces useful results and stays on time and on budget, every data science team needs a project manager. No matter how big the data science team is, the role of the project manager should be firmly planned in the team from the start.

Which skills should the first data scientists bring to the team?

Here, too, there is no generally applicable rule. Ideally, the first team members have broad methodological skills and have practical experience in project implementation. In addition, the first data scientists should have a good understanding of the industry and the business processes in their company. Because the first team members in particular will work a lot with decision-makers and employees from different departments.

At the beginning, the data science team will be heavily involved in the selection of suitable use cases. The expertise of the data experts is also required when interpreting data and recommending recommendations for action. Ideally, the first data science initiatives are tested in the form of pilot projects. It is best to start with smaller projects that are as easy to implement as possible. If these were successfully implemented and brought into a productive setting, companies can devote themselves to more complex issues. For this it can make sense to recruit data scientists with special know-how in one area. At a later point in time, specialists complete the more generalist team from the early days.

How should the team be integrated into the company?

Sometimes it makes sense to place a data scientist with special domain knowledge directly in a specialist area, such as controlling or marketing. In such a case, there is no dedicated data science team. The disadvantage of this, however, is that data experts can rarely talk to their direct colleagues about statistical processes. The exchange between like-minded people from other departments must therefore be promoted in a targeted manner.

Much more often, data science teams are set up that serve a wide variety of specialist areas in the company. Thus, the data science team acts as an internal service provider. The big challenge lies in the communication between the technical data scientists and the departments from which the specifications for data science projects usually come. A good understanding of the work of the other department is essential for the success of initiatives and the data science team.

An additional danger is that data science teams will become silos. Valuable knowledge only stays within the team and cannot be used properly elsewhere in the company.

If you want to read more about how to successfully build a data science team in a company, you can find my guest post on malt in full HERE. Have fun reading