In the last few years there has been an explosion in the number of courses and programs to train the next generation of data scientists. This is in response to both the perceived demand for data science (along with the rest of the buzzword soup: Big Data, the Internet of Things, Machine Learning, AI, etc.) and the perceived shortage of trained data scientists. While finding good data scientists is still a difficult problem, the increasing number of bootcamps, master’s degree programs, and online courses is starting to address this difficulty. In our experience, however, we find that many organizations are not necessarily short on data science talent, but rather that management does not properly use the talent they already have. Managers, directors, and VPs must improve their ability to work with data scientists if they want their organizations to use data more effectively; in short, they need to be more data fluent.
While the structures of organizations are complex and highly variable, this diagram captures some of the key roles in an effective data-driven organization:
The roles between senior leadership and data scientists fulfill the critical function of translating—translating business strategies into tangible initiatives, framing projects appropriately for data scientists, interpreting and critically examining results from data scientists, and communicating those results back to senior leadership in a compelling way. In order to translate, one must be fluent in both business and data. We at Datascope feel that this translation process can be broken down into the following four steps: Clarify, Iterate, Interpret, Communicate.
Clarify: Abstract Strategy to Concrete Projects
Many companies have started initiatives to become more “data driven”; the issue here is that “becoming more data driven” is such an ambiguous problem that it’s often difficult to know where to start. While senior leadership identifies the big-picture goals for a business, data science managers need to come up with concrete steps to help achieve these vague strategic goals. A good way to craft well-defined projects is to borrow processes from the design community and treat problem definition as a stage in the project where true needs are emphasized. From there, one synthesizes ways to define problems that, when solved, meet most or all of the needs. It’s hard to resist the temptation to look at available data and imagine what can be done, but a good data science manager starts from the end point, focuses on designing an experience that aligns with big-picture goals, and only then assesses how to technically achieve it.
Iterate: Managing the Data Science Process
It’s not necessary to actually be a data scientist to manage data science projects. It is crucial, though, to have an awareness of the difficulty of various data science tasks so that projects can be effectively planned. For example, it’s important to know that setting up a process to gather top ten lists from a website is a straightforward task that can likely be done in a few hours. However, some tasks that sound easy enough are deceptively difficult, as the XKCD comic to the right points out.
Because tools and techniques can sometimes evolve rapidly, staying aware of the common data science tasks and how they can be accomplished is not easy. In the face of so many options, one of the best ways to keep a project moving forward is to get comfortable with fast prototyping, iteration, and pivoting quickly toward methods that best meet the needs of senior leadership.
Interpret: Consuming Your Data Responsibly
Interpreting data-driven results, despite the air of objectivity, has always been tricky. Reports that include numbers, charts, and correlations can lead to acceptance of results as scientific even when there are critical errors. For VPs, directors, and managers, it’s critical to be able to question data-driven results; this involves understanding what to trust, knowing what “smells,” and ultimately making the best-informed decisions from the data.
A healthy skepticism is a manager’s best friend when interpreting data-driven results. Data science managers need to know what questions to ask, become aware of common problems, and learn some fundamental statistics that will allow them to diagnose those problems. Managers can build a powerful B.S. detector with a solid but basic statistical foundation, without the full background of a data scientist.
Communicate: Sending Results up the Chain
A bad outcome of a data project is to execute it well, but then fail to properly convey your success back to senior leadership. Data science managers need to know the best ways to effectively communicate quantitative information. Again, it’s not necessary to be an information designer or an Edward Tufte disciple, but make sure to stay on top of the basic skills for telling a compelling story driven by quantitative results. At this stage in the game, it’s also crucial to keep in mind that senior leadership isn’t as deep in the weeds as data scientists are. Data science managers need to help the higher-ups see the big picture as quickly and plainly as possible, guiding them to the “so what?” behind the project without glossing over any important information.
DS4MAD is Officially in Session!
By clarifying ambiguous strategy into tangible data projects, managing those projects iteratively, interpreting the results carefully, and communicating project significance clearly, VPs, directors, and managers can translate between the C-Suite and the data geeks to help an organization become data fluent. We here at Datascope feel so strongly about this that we’ve designed a two-day course to help middle managers master these four skills and lead data scientists more effectively. It’s called Data Science for Managers and Directors (DS4MAD) and it’s coming to the Central Loop on November 16! To learn more about turning your company into a well-oiled data-driven machine, click here and sign up for DS4MAD!