Recovering $35 million in lost sales using machine learning

Insurers lose sales inside conversations that no dashboard can see. One of North America's largest insurers wanted behavioral science built into its digital and customer experience strategy, not bought project by project, and asked us to design a permanent in-house team and prove its value on a first project.

We designed the team from evidence: a review of the client's internal processes and products, one-on-one interviews with 25 operational stakeholders, and direct input from advisors who had built behavioral teams at the White House and Google. The result was an organizational blueprint, hiring profiles combining behavioral science with digital expertise, and a hired team.

For the first project, we ranked candidate pilots by return, feasibility, and spillover, and selected agent sales calls. We analyzed more than 170,000 hours of calls, using process mapping to locate the critical moments in each conversation and machine learning to cluster where and how calls go wrong. From those clusters we built targeted script changes and agent training, then piloted them against a control group.

Sales rose 12% from the first pilot alone, against the control group, with a projected $35 million increase in annual revenue. The larger asset is permanent: the insurer now runs its own behavioral science team, structured, staffed, and proven on a live commercial problem.