This week, I had the distinct honor of chairing the first annual Data Science Leaders Network East Coast meeting in Boston, Massachusetts. I was privileged to be joined by the top data science leaders from some of the largest, most innovative, most profitable organizations not only in their respective industries, but in the world – including leaders from Google, Uber, Coca-Cola, Capital One, BCG, Wayfair, and others.
The DSL network was developed based on the need for a professional network for leaders in this new but growing role. My biggest realization from the meeting was that as talented, insightful, and forward-thinking as many data science leaders are, the fact remains that data science is still very much a nascent enterprise. As data science leaders, many of us are still finding our feet — not in terms of the mechanics and technicalities of our daily work, but in terms of carving out our niche within both our individual organizations and the global economy at large.
It’s clear that data science is revolutionizing business practices across every sector — there’s no question about that — but before we move forward as a discipline, data science leaders must start to address some of the challenges already at hand. There are many worthy of mention — but I’d like to briefly touch on three key takeaways that struck me as particularly relevant.
Key Takeaway #1: Alignment of Data Science with the Right Business Objectives
First, as senior leaders in the data science space, it’s our responsibility to ensure that our teams’ efforts are aligned with the right business objectives. Whether we’re working on improving our own internal operations or delivering analytics-as-a-service to our clients, whether we’re working in the financial sector or the industrial sector or the retail sector, our North Star should always be the same: tangible business value.
For the technicians and academics among us, it’s easy to find evidence of a job well done not in the achievement of a predefined business outcome, but in the execution of a highly complex analytics program itself. Tragically, the intricacy of our algorithms is immaterial unless it drives real world results.
While its name might suggest otherwise, there’s a certain artistry to data science, at least where its application is concerned. Just as the best brushes and the finest canvasses do not a master painter make, the largest datasets and the most sophisticated algorithms are no guarantee of an effective data science operation. As important as the appropriate application of machine learning and statistical inference may be, at the end of the day, they’re just tools.
In my experience in healthcare marketing, the most successful data scientists are those who pair superior tools — comprehensive datasets, well-crafted algorithms, and so forth — with an aptitude for leveraging those tools in pursuit of improving patient’s lives in a clear and strategic fashion.
Our first priority as leaders should be to help our teams wield these tools as artfully as possible. Only then will our efforts create enough tangible value to be taken seriously by our business leader counterparts.
Key Takeaway 2: Moving Beyond Buy-In, and Integrating Data Science into Business Culture
It takes a specific set of conditions to facilitate meaningful, sustainable implementations of data science. One-off initiatives layered over an organization’s existing processes might create a veneer of data science, but real change only occurs once data science has been spliced into an organization’s DNA.
This demands what I call a “culture of analytics enablement,” a culture in which data science is recognized — and supported — as a key component of an organization’s core business processes. This means several things for leaders in the data science space.
Most importantly, we need to break down silos within our organizations. The best data science team in the world will struggle to drive results if it encounters barriers to cross-departmental communication and collaboration at every turn. In practical data science, isolation is the enemy of innovation.
It’s been made clear to me through learning from data science leaders across other industries: dismantling organizational silos is often easier said than done. Doing so requires data science leaders to wear many hats. During one meeting, we have to be teachers, guiding less data-focused colleagues through the basics of our initiatives. During the next, we have to be leaders, presenting a unified vision around which our data science talent can structure their day-to-day work. During a third, we have to be spokespeople (and sometimes cheerleaders), communicating to clients or investors how our data science initiatives set us apart from our competition.
Weaving data science into the fabric of the organization entails finding answers to difficult questions. Are initiatives designed to scale alongside your organization’s business? Are they future-proof, at least as much as possible in a field evolving as quickly as data science? Are they not only legally compliant, but ethical, as well?
These are questions business leaders grapple on a day to day basis, and as data science leaders, we must be prepared to contribute to these conversations if we are to stake a claim to an authority extending beyond our own departments.
Key Takeaway #3: Winning the Data Science Talent War
Finally, data science is as much about people as it is about numbers. The only problem is that there aren’t enough right-skilled people to go around.
Simply put, we find ourselves in a labor market where demand has vastly outpaced supply. According to a report published by LinkedIn in December, three of the top five fastest growing jobs in the United States are directly related to data science. Opportunities for machine learning engineers increased nearly tenfold between 2012 and 2017. Opportunities for data scientists increased by a factor of 6.5. Opportunities for big data developers increased by a factor of 5.5. And, much to our collective benefit, opportunities for directors of data science increased nearly fivefold.
Unfortunately, the report also points out that while opportunities for data scientists have skyrocketed by 650 percent since 2012, a mere 35,000 U.S. residents currently have a robust data science skill-set.
We all need strong, diversely-skilled teams, but building one can sometimes feel more like searching for a needle in a haystack than conducting a simple recruitment push. And to be frank, beyond forging partnerships with colleges and universities aimed at funneling more students into data science, there’s little we can do to deepen the talent pools from which we draw.
As such, we need to make a concerted effort to create workplaces where the few data scientists that are out there want to work. That means articulating a clear organizational mission. It means valuing input from junior team members. It means creating a safe and supportive work environment. And, in some cases, it means paying a premium to get the person you really want.
Ultimately, events like the DSL network enable me to confidently say that we as a profession will rise to meet these challenges. Each of the challenges we face is an opportunity — for reflection, for improvement, for community-building across the growing Data Science community.