By David Shilane, Lecturer, M.S. in Applied Analytics, Columbia School of Professional Studies
These days, nearly every company aims to build a data-driven organization. In practice, this simple goal often turns out to be elusive or only partially realized. Managers expect that data systems and analyses will lead to direct improvements in performance. Likewise, when data scientists make recommendations based on statistical models, we assume that the implementation will be straightforward. Unfortunately, driving organizational changes is not so simple. To provide some examples from my professional practice: In healthcare, organizations I have supported have designed programs to prevent hospital readmission for high-risk patients. In non-profit management, I have studied fundraising campaigns aimed at increasing contributions from existing donors. In these cases, the strategic choices and program designs proved to be challenging. This is where data science meets workflows.
My Research Design course emphasizes organizational goals. I like to ask my students to build a managerial achievement from the ground up. What represents success for one medical patient or one potential donor? Across the population, we calculate a performance metric to quantify the organization’s overall success for each goal. So we could track the percentage of patients who avoid hospital readmission or the average yearly donation. Then I ask my students how we might improve performance. Strategic initiatives might include more timely outreach to medical patients, purpose-driven fundraising campaigns, or any number of other alternatives. With our Research Design playbook, we can experimentally test new strategies in real-world settings. Quantitative methods can estimate the effectiveness of new strategies and assess their statistical reliability. Then, we can deliver informed recommendations to the management. But that is not the end of the story!
Updating product designs, business services, or organizational structures is rarely straightforward. Implementing new strategies requires training, systems, and a barrel of patience. In my professional practice, I have observed this transitional process across a range of companies. In my experience, driving change is best supported by engineering systems that integrate data collection, analysis, and workflows. For healthcare organizations, this means assigning tasks to care providers and case managers, then tracking their checklists of follow-up care activities. For non-profits, workflows involve scheduled outreach activities to ensure that each potential donor is contacted, acknowledged, and informed about the donation’s impact. Monitoring progress helps to ensure consistency and timely execution. Better yet, workflow systems generate new sources of data to analyze. Now we can not only assess the quality of a new strategy but also understand and address the barriers to successful implementation. Improvements in process metrics may lead to improvements in outcomes, financial success, and overall performance. Statistical modeling can help us understand and quantify these relationships in their proper context.
The confluence of data science and workflows demonstrates the overall importance of integrating technology and management. It also creates a compelling argument for studying both subjects in our Applied Analytics master’s program (and the Research Design course). Ultimately, combining data science and workflows drives meaningful improvements in business processes, bringing us ever closer to fully-realized, data-driven organizations.
About the Program
Columbia University’s Master of Science in Applied Analytics prepares students with the practical data and leadership skills to succeed. The program combines in-depth knowledge of data analytics with the leadership, management, and communication principles and tactics necessary to impact decision-making across industries and organizational functions.
Learn more about the program here. The program is available full-time and part-time, online and on-campus.