Morgan Stanley projects that AI can reduce the insurance expense ratio by 200 basis points by 2030, from 30.5 to 28.5 for the P&C sector. WTW's March 2026 survey found that insurers using more sophisticated analytics achieved combined ratios six points lower than slower adopters between 2022 and 2024.
For a carrier with a billion-pound premium portfolio, 200 basis points of combined ratio improvement is 20 million pounds of additional underwriting profit. The question for boards is not whether AI improves margins. It is where the improvement comes from and how to govern the transition.
The Combined Ratio in 2026
The combined ratio is expected to worsen from 97.2% in 2024 to approximately 99% in 2026, driven by slowing premium growth and rising claims costs. This structural headwind makes the AI margin opportunity more urgent, not less. Carriers that cannot offset cost pressures through operational improvement will face margin compression that market conditions alone will not resolve.
The 200 basis points that AI can deliver come from three sources, in roughly equal proportions: underwriting accuracy, claims efficiency, and expense reduction. Each requires different investment and different governance.
Source 1: Underwriting Accuracy (60 to 80 Basis Points)
Better risk selection and pricing is the oldest promise of AI in insurance, and the most proven. Carriers deploying AI-enhanced underwriting report loss ratio improvements of three to five percentage points, though results vary significantly by line and by the quality of the underlying data.
The mechanism is straightforward. Traditional underwriting uses a limited set of rating factors processed through generalised linear models. AI models can incorporate a wider signal set (telematics data, satellite imagery, behavioural patterns, external data sources) and capture non-linear relationships that GLMs miss. The result is more accurate risk pricing: fewer under-priced risks that generate losses, and fewer over-priced risks that drive away good business.
The board-level question is whether this accuracy advantage is sustainable. Three considerations:
First, as more carriers adopt AI underwriting, the competitive advantage shifts from "having AI" to "having better AI." This means the investment is ongoing, not one-off. A carrier that builds an AI underwriting model in 2025 and does not iterate will see its advantage erode as competitors catch up.
Second, underwriting accuracy depends on data quality. Carriers with proprietary data assets (large portfolios, telematics programmes, connected-device partnerships) have a structural advantage. Those relying on the same third-party data as everyone else are building models that converge to the same answers.
Third, regulators are watching. EIOPA's August 2025 Opinion on AI Governance requires that AI pricing models be fair, explainable, and free of bias. Underwriting accuracy that comes at the cost of fairness is not sustainable margin improvement; it is a conduct liability. Boards need to see fairness testing results alongside loss ratio improvements.
Source 2: Claims Efficiency (60 to 80 Basis Points)
Claims handling is the largest operational cost in insurance, and the area where AI delivers the most immediate returns. Early adopters report 40% or greater reductions in claims cycle times, with straight-through processing rates increasing from 10 to 15% to 70 to 90% for simple claims.
The margin improvement comes from three mechanisms.
Faster settlement reduces claims inflation. A motor claim that takes 60 days to settle incurs hire car costs, storage fees, and customer frustration that drives complaint costs. The same claim settled in 10 days costs less, produces a better customer outcome, and reduces the insurer's expense load.
Fraud detection improves leakage. AI fraud detection systems report 30% or greater improvement in detection rates over rule-based systems. For a carrier with a 5% fraud rate on a 500 million pound claims portfolio, a 30% improvement in detection is 7.5 million pounds of avoided leakage.
Automated triage reduces handling cost. When 70% of simple claims can be triaged and processed without human intervention, the claims operation can be restructured. The headcount saving is real, but the more significant margin impact is that human adjusters spend their time on complex, high-value claims where their expertise adds value, rather than on routine processing.
The board governance requirement for claims AI is outcome measurement. Not "we implemented AI in claims," but "AI claims automation delivered X pounds of margin improvement in the half, comprising Y from faster settlement, Z from fraud reduction, and W from handling cost reduction." If the team cannot produce this decomposition, the board cannot validate the investment.
Source 3: Expense Reduction (40 to 60 Basis Points)
The expense ratio in insurance includes distribution costs, administrative costs, and the overhead of the operating model. AI contributes to expense reduction across all three, but the magnitude depends on how aggressively the carrier restructures its operations around AI capabilities rather than simply layering AI onto existing processes.
Underwriting expense ratios are projected to decline 15 to 20% in P&C and more than 25% in life insurance, according to industry estimates. These projections assume that carriers redesign underwriting workflows rather than using AI as an addition to existing processes.
The distinction matters. An insurer that deploys an AI underwriting assistant alongside its existing underwriting team may see accuracy improvements but limited expense reduction. An insurer that redesigns its underwriting process so that AI handles 60% of submissions autonomously, with human underwriters focusing on complex risks, achieves both accuracy and expense benefits.
The board decision is an operating model decision, not a technology decision. How much of the underwriting, claims, and administrative operation should be AI-native versus AI-assisted? The answer determines the expense ratio trajectory.
Governance for the Board
In my work with carrier boards, the most common governance gap is the absence of a clear attribution framework connecting AI investment to combined ratio movement. Without it, AI spending is defended with anecdotes and the board has no basis for deciding whether to scale, sustain, or wind down a programme. A CFO I advise recently presented AI margin attribution using the same decomposition framework the board applies to reinsurance decisions. The discipline of separating underwriting accuracy gains from claims efficiency and expense reduction forced the executive team to confront where AI was genuinely delivering and where projected savings had not yet materialised.
For boards seeking to govern the AI margin opportunity, four metrics matter.
AI-attributed margin improvement. What portion of the combined ratio improvement in the period is directly attributable to AI deployments? This requires a measurement framework that isolates AI impact from market conditions, portfolio mix changes, and other factors.
Investment-to-return ratio. What has the carrier spent on AI (total cost of ownership including infrastructure, talent, governance, and regulatory compliance) versus the margin improvement delivered? The ratio should improve over time as foundational investments amortise and AI applications scale.
Risk concentration. How dependent is the carrier's margin on AI systems? If a key model fails or is withdrawn by a regulator, what is the impact on the combined ratio? This is not a reason to avoid AI. It is a reason to ensure that AI risk is on the risk register, with appropriate contingency plans.
Fairness and conduct metrics. Is the margin improvement coming from genuine operational efficiency and underwriting accuracy, or is any portion attributable to practices that would not survive regulatory scrutiny? The board needs assurance that the 200 basis points are clean.
The carriers that will capture the full AI margin opportunity are those that treat it as a board-level strategic programme, not a technology initiative. The combined ratio is the board's number. AI is the lever. The governance framework must connect the two with the same rigour that the board applies to every other material driver of financial performance.
*To discuss how the 90-Day AI Acceleration programme can help your board capture the AI margin opportunity, contact the Value Institute.*