The Gap Between Spending and Results
BCG’s AI Radar shows property and casualty insurers will triple AI spending as a share of revenue in 2026, yet only 38% are generating value at scale from AI in core workflows like underwriting and claims. BCG frames this as a strategic gap rather than a technical one: AI dropped into legacy operating models built for human-led execution simply doesn’t pay off, regardless of spend. The value on the table for those who get it right is substantial. Early adopters are on a path to 3-5% premium growth, worth an extra 80 billion dollars in the US alone, plus 15-25% reductions in operating cost per premium dollar, worth 35 to 60 billion dollars nationally.
Why This Extends Past Insurance
The pressure BCG describes in P&C, soft pricing, rising claim severity, reinsurance capacity strained by climate risk, is industry-specific, but the underlying diagnosis applies directly to collections and lending operations too. Insurance already leads most other sectors in AI pilots, yet has one of the weakest records at converting that experimentation into scaled value, because many insurers focus on the technology without a strategy to capture value across the whole process. A collections operation that adds a voicebot on top of an unchanged workflow is repeating the same mistake BCG is describing in claims: the tool works, but the process around it hasn’t changed, so the value never fully materializes.
The Detail Most Teams Get Backwards
BCG’s resourcing framework, the 10-20-70 model, is the part worth sitting with longest. Only 10% of a successful AI transformation’s effort goes into the algorithms themselves. Another 20% goes into technology and data, including building a shared ontology so AI can interpret structured and unstructured information consistently across the business. The remaining 70%, the majority by far, goes into building an agent-first operating model: talent, culture, change management, and governance. Most organizations invest in roughly the reverse ratio, treating the model as the hard problem and the surrounding operating model as an afterthought, which is a large part of why the returns don’t show up.
What This Means for Collections Operations
BCG’s three linked plays, Deploy, Reshape, Reinvent, give collections leaders a useful way to locate where they actually stand. Deploy is the shallow layer: using AI to summarize submissions or guide agents in real time without touching the underlying workflow. It frees capacity but doesn’t change the core economics. Reshape is where the real value sits: redesigning the end-to-end process so AI agents handle execution directly while humans step in only when exceptions require judgment, the same distinction BCG credits with cutting time-to-quote by 30-40% in insurance intake. Reinvent goes further still, using AI to change the underlying business model itself. For most collections operations, the immediate opportunity is closing the gap between Deploy and Reshape, since that gap is exactly where BCG’s data shows the 62% of organizations without scaled AI value are currently stuck.