Can insurers turn AI pilots into a profit engine?
Distribution costs and weak AI integration are limiting insurers’ returns.
Insurers risk missing the biggest financial gains from artificial intelligence (AI) if they keep the technology in isolated pilot projects, with distribution costs consuming as much as $0.80 of every first-year life premium, according to McKinsey & Company, Inc.
Global gross written premiums have grown 4.9% a year since 2005 to about $8.3t in 2025, whilst profits before tax increased 4.3% annually to about $580b, McKinsey said in a July report.
The gap points to weaker operating efficiency despite steady premium growth. Premiums as a share of global economic output have also remained flat for a decade, even as the natural catastrophe protection gap reached $133b in 2025 and less than 1% of global cyber costs were insured, McKinsey said.
Commissions account for $0.10 to $0.25 of every premium dollar in property and casualty insurance, whilst insurance cost ratios are 17% higher globally than in 2005, it said.
“The real question for insurers isn’t whether to invest in AI; it’s how to embed it into underwriting, claims and growth in a way that’s governed, scalable, and built to last,” Bruno Abril, NTT DATA Group Corporation’s global lead for the insurance industry, said in a June report.
NTT Data found that 85.8% of insurers whose AI strategies were fully aligned with their business strategies reported profit increases of at least 5% from AI, compared with 45.5% of insurers with less-aligned strategies.
AI leaders are also spending more. About 71.7% described their AI investment as “very significant,” compared with 51% of lagging insurers, whilst 63% planned to increase spending further.
NTT Data surveyed 291 insurance executives and classified 46, or 16%, as AI leaders.
McKinsey said AI could reshape how insurers sell and service policies, including through AI assistants that compare coverage and switch policies on behalf of customers.
AI has also reduced onboarding costs by as much as 40% and increased agent productivity by 10% to 20%, it said.
More than half of AI leaders are rebuilding core underwriting, claims, and policy systems with AI embedded into them, compared with 6.5% of lagging insurers, NTT Data said.
Governance remains another dividing line. About 67.4% of AI leaders use centralised AI governance, compared with 23.5% of lagging insurers, whilst almost 85% have a dedicated chief AI officer, compared with 45.1% of lagging insurers.
“AI could upend the industry's long-standing structures and dynamics,” McKinsey said.
Questions to ponder:
- Can insurers capture AI’s cost savings without weakening customer service or underwriting quality?
- Will AI reduce distribution costs enough to materially improve insurance profitability?
- How much centralised oversight is needed before insurers can scale AI safely?
Read Here
Advertise
Sign Up
EXPERT OPINION
AI will become an impact engine for insurers only when it moves beyond isolated pilots and is embedded across distribution, underwriting and claims etc. It certainly presents a significant opportunity for insurers to strengthen distribution, sharpen underwriting, and deliver personalised claims experiences. Scaling responsibly - with strong governance and human oversight - will be essential to preserving trust.
AI is increasingly being deployed across insurance operations, with applications spanning virtually every function. The industry broadly expects AI to improve process efficiency, strengthen risk assessment and decision-making, and enhance data quality and collection—ultimately supporting greater profitability. As AI evolves into a driver of profitable growth, it can enable insurers to improve risk selection, reduce claims leakage, increase conversion and retention rates, and lower the overall expense base. Realising these benefits will require more than technology investment; insurers must also redesign workflows, strengthen data foundations and collection, establish effective governance, drive user adoption, and maintain clear accountability. At the same time, AI must be supported by proportionate governance and controls covering risk registration, outcome validation, usage monitoring, issue escalation, data privacy, and ongoing regulatory compliance.
As insurers think about transforming distribution through AI, the opportunity should be viewed first as a growth lever, not simply a cost lever. An example of this in agency and bancassurance channels would be to use agentic AI-powered conversational platforms deployed across frontline sellers and managers. The platform could put knowledge and intelligence directly in the hands of the frontline by improving access to product information, enabling the generation of personalized sales and marketing material for customer communication, and supporting other day-to-day sales activities.
More broadly, transformations of this kind can enable a fundamental shift in the sales operating model. AI can codify the practices and knowledge of top performers and make them available at scale, helping sellers and managers become more self-sufficient. Over time, this can reduce reliance on multiple layers of sales supervision, expand spans of control, flatten parts of the sales hierarchy and lower distribution costs while sustaining, or improving, frontline productivity.