What are data scientists?

“We’re not looking for the overly complex analyses of a data scientist; we just need someone who can simply deliver us clear insights that help us understand our business better”. It was the comment I received recently when pitching for a position at a large company that was in search of a data-driven marketer. The statement caught my attention, as it conflicted with my personal perception of a data scientist. Up until that point, I had always associated the term ‘data science’ with one’s ability to turn raw data into valuable business insights. So I asked myself: isn’t helping companies understand their business thoroughly the exact function of a data scientist? Clearly, not all managers seem to think so.

What’s a data scientist anyway?

The point is that in today’s business world, there seems to be a gap when it comes to the way people look at data scientists. Some argue that data scientists are destined to change the old way of doing business in pretty much every way possible. Others react more tempered and dare to put their usefulness in question. That is why I decided to do some research. More specifically, I wanted to find common ground between what makes a data scientist and what she or he actually does.

According to SAS, one of the long-time leaders in software solutions, a data scientist is someone who “digs in and unearths business insights that no one thought to look for before”. This means that in its description of a data science profile, the company puts its emphasis on the scientist’s active contribution in the field of business intelligence. Although often competed by SAS, IBM reconfirms, stating that the primary focus of data scientists should be on their business objectives and not on the technology itself. Taking a more scholarly approach, Harvard Business Review speaks of data scientists as the people who know how to make good use of data in order to answer critical business questions. You get the point. Although often differently formulated, both practitioners as well as academics seem to agree on the idea that data scientists should primarily focus on creating business value.

The good, the bad and the in-betweens

Yet, if the consensus is that data scientists are there to make life easier for the management, then why are many executives skeptical of hiring data scientists in a business context? That is because in practice, a lot of – if not most – data scientists do not follow the classical description of their job. I believe that there exist three groups of data scientists.

The first group consists of those who work according to the ‘your wish is our command’ principle. Without questions asked, these scientists focus exclusively on carrying out a technical analysis or developing a certain data product with the utmost precision from start to end. The interpretation in terms of deployability however is left to a third party. According to the classical definition of a data scientist however, one could say that these are bad data scientists.

Next, you have the group of in-betweens. These data scientists make sure that once their analysis is completed or their product is finished, the resulting insights are clearly reported to stakeholders. In this way, they are able to create value by offering an answer to the most pressing questions and needs of the management.

Yet, good data scientists even take it one step further. Besides carrying out a great data project and reporting it to the management, they dare to challenge the central business question up-front. More specifically, together with the requesting party, these scientists proactively think about the relevance of the question and its impact on business figures.

Proactive evaluation

Take for example the common task of predictive analysis. Nowadays, a data scientist can predict practically any possible consumer behavior. The question is however, if this predictive model is relevant enough to contribute to the predetermined business goals of a company.

For instance, one can predict the amount of blue shirts a shop will sell in the upcoming month. If blue shirts generate 95 percent of the company’s turnover, then in terms of target setting this could be a good idea.

But what if the prediction of blue shirts turns out to be irrelevant for our business? Maybe, if we are a shirt company, we could upscale the analysis to other colors of shirts such as green and yellow? Maybe, if we are a more general e-commerce, there are other product categories that are more crucial to the sales figures than shirts? Even more so, other types of predictive models such as churn or customer lifetime forecasting could possibly have a bigger impact on turnover rates.

These, amongst many more, are all questions a good data scientist would ask before she or he proceeds to the development of a data project.

Answering the question

In conclusion, we wanted to find out what data scientists are to your business, friends or foe? Well, it depends on which group of data scientists they belong to: the good, the bad or the in-betweens. Data scientists at The Reference strive to be part of that first group. That is why we adopted the term data consultant for all of our team members. Although data consultants possess the full range of technical capabilities of a data scientist, the emphasis is put on the bigger picture: the concrete guidance throughout a data project that enables managers to make well-informed decisions.

Are you interested in staying up-to-date with our data science insights?



Don't miss out

The Reference has its office in the heart of Manhattan.
“I want to wake up in that city that never sleeps, and find I'm king of the hill, top of the list, head of the heap” – Frank Sinatra