Allstate’s Data, Discovery, and Decision Science group (D3) had started as a small team of data scientists helping to quantify insurance risk accurately, but by 2017, it had ballooned to 300 employees offering data capabilities to the firm’s business verticals (e.g., Claims or HR). Allstate’s senior leadership team had recently defined the strategic objectives that should be driving D3 projects, but it quickly became apparent that D3’s current structure was incongruent with these strategic goals. D3 members and leadership agreed: A reorganization was necessary. However, an earlier reorganization, which had reassigned employees without their input, had left them feeling burned. Senior leaders didn’t want to make the same mistake twice but knew they didn’t have time to post and review applications for each position in the new structure.
Given that D3 members were data nerds, the leadership team decided to bet on their own data talent by developing and implementing an algorithm to determine which position each manager and employee should fill. This algorithm, designed by one of their own engineers with the goal of “maximizing employee happiness,” allowed for objectivity, employee feedback, and an acceptable timeline. Despite leaders’ attempts at total transparency and clear communication, D3 members encountered some system-gaming, dissatisfaction, and anxiety throughout the process. This case presents an unbiased view of the realignment and prepares students to debate the success of using AI in talent management, as well as discuss how to manage such a process effectively.