The widening spectrum of data science roles

The development and use of self-service analytics has brought with it a new role in many organisations: the citizen data scientist. But is this genuinely a new role, or is it just a new name for a business analyst?

Business analysis is broadly defined as analysing the business, including processes or systems, and putting forward solutions. I would say that all those were part of the citizen data scientist’s role, which suggests that it may be the term alone that is new. But I think there is one key difference: citizen data scientists have to do all these things, but these functions are not at the core of that person’s role.

In other words, citizen data scientists are not actually data scientists. Instead, they are business users who understand and can do analytics. They are able to use data but that is not their primary role in the business. Uniquely, they bring together knowledge of the business, and some understanding of analysis. The best citizen data scientists know the business, and are not afraid to get their hands dirty with data. They use the right tools to generate insights that can bring value to the business. It is, however, up to the business to evaluate the insights.

The role itself has arisen as we all become more data-driven. At the same time, a shortage of dedicated data scientists has meant that there is a lack of capacity to crunch numbers and produce insights. In other words, citizen data scientists have developed out of necessity. It is, perhaps, best described as a hybrid role, but is certainly filling a gap in many organisations.

But as citizen data scientists are not dedicated data scientists, they need more support to enable them to perform their role successfully.

First of all, they need suitable self-service analytics tools to enable them to manipulate data and achieve information and insights. These tools need to be simple to use, but with a wide range of analytical options available. Citizen data scientists may also need training and support in using these tools, at least initially.

The next requirement is suitable data. This means from multiple sources, because insights seldom come from rehashing a single source, and the data also need to be clean and high quality.