The flip answer is yes of course you need a data scientist! The serious answer is more nuanced. When a term gets hot, it can easily get overused and generate backlash. Here is my quick check list of cases that might warrant an investment in data science.
Data Products: If your organization sells data you probably already have many of the characteristics of data science even if you don’t have someone with that official role. A data scientist can help you with everything from algorithmic anomaly detection, to measurement validity and explore how to move up the value chain. Raw data gains value through segmentation, clustering, time-series historical comparisons and prediction of upcoming value.
Business Arguments: If your executive team is fraught with arguments about the fundamental assumptions and hypothesis underlying the business, you may need a data scientist. If might just feel like an argument to you, but someone trained as a scientist can translate those arguments into research questions and begin to define what data counts as evidence. They won’t be able to remove all uncertainties, but imagine your organization changing the basis of decisions from fighting fires and incident management into an iterative a series of experiments generating increasing confidence in the empirical foundation of your business.
Service Acceleration and Automated Personalization: Business process automation has gone through several iterations over the past 40 years. Today we are moving toward increasingly automated service delivery, scheduling and personalized offerings. Mobile devices and social media combine to provide businesses new kinds of location, time and interest data with which to design custom personalized JIT (just-in-time) experiences and offers. If your business is moving in this direction you need a data scientist or two to help you get the automated analytical components designed correctly and to monitor the ongoing effectiveness of the implementations.
Scale: Organizations with billions in revenue can easily justify the use of data scientists. An improvement of 1% on $100 million in profits easily pays for a data scientist for the year. In many organizations there are much larger inefficiencies to be found and it doesn’t take a scale of billions in order to pay-off. At the other end of scale are many tech start-ups in Silicon Valley and elsewhere are hiring one or more data scientist right from the beginning as a competitive advantage.
Data Tsunami: If your organization is struggling with Big Data you realize it. If your organization has a sense of drowning in data and yet still lack true insight you need help from a data scientist. A data scientist can help you connect the silos, separate signal from noise and help you get a handle on which levers can really move the business.
Intuition: “I just have this feeling that we aren’t getting everything we should out of our data?” One of my longest running engagements started just this way. We started with a short project and after about a week’s worth of work it morphed into an ongoing engagement to help the team a few hours per week with data exploration, predictive analyses, exploring competing business hypotheses and testing the effectiveness of promotions. Scale for this line of business was initially quite modest, but getting the new effort grounded in data early was a big win. This client now runs a significant part of their business in a highly automated fashion with market responsive pricing. And, they were able to recover a portion of the data science costs out of licensing data subsets to non-competitive industry partners.
Does your organization need a data scientist? If you found yourself nodding in agreement with any of these cases you might.