BCG's 10-20-70 Rule for Health Insurers Puts People Ahead of the AI Itself

BCG is telling health insurers that off-the-shelf AI tools deliver 10-20% efficiency gains per employee, while agentic AI applied to fully redesigned workflows can deliver 30-80% — a gap wide enough to reframe the adoption decision entirely. The firm cites a European insurer where agentic AI cut claims-processing time from 30-60 minutes to 2-4 minutes, and credits automated fraud detection with reducing claims leakage by 3-5% of total payout. One client relationship saw customer retention gains of up to 14 percentage points alongside claims-cost reductions of up to 70%. BCG’s stated reason for the industry’s urgency: insurers stand to lose 30-50% of their experienced workforce to retirement over the next decade.

The number worth flagging for anyone advising on this is BCG’s own prescribed effort split — the “10-20-70 rule” — 10% of implementation effort on the AI application itself, 20% on technology and data infrastructure, and 70% on people, organization, and process change. That ratio is an admission, coming from a firm selling AI transformation work, that the technology is the smallest part of what determines whether the efficiency numbers above actually materialize.

Bain’s July insurance outlook is the reality check that belongs next to this. Bain found direct written premiums had doubled over ten years while expense ratios fell only one percentage point — a stark mismatch between AI spending and measured productivity gains, even as hiring in the sector is down nearly 50% since 2022. BCG’s case for what agentic AI can do and Bain’s data on what AI spending has actually produced so far aren’t contradictory; they’re describing the same gap from opposite ends. For consultants advising insurance and healthcare clients, the honest framing isn’t “AI delivers 30-80% efficiency gains” — it’s that the gains BCG describes require the change-management work most insurers haven’t yet done, which is exactly what Bain’s numbers show.