How Claims-Processing AI Helps Insurers Reduce Leakage to Improve Combined Ratio

An insurer’s margins depend on their combined ratio. It’s the clearest measure of your core insurance operation’s performance.

Carriers know there are only two paths to increase profitability: increase premiums or reduce expenses & losses. The former is becoming less viable, given regulatory constraints, slowing growth, and customer attrition.

However, AI for insurance claims opens a sizable opportunity to reduce both losses and expenses, helping insurers mitigate fraud losses and optimize operational efficiency.

In this article, we’ll explore:

• What combined ratio is and how it’s calculated;
• Limitations of traditional approaches to improving combined ratio;
• How AI reduces both loss ratio and expense ratio; and
• Why deterministic AI matters for sustainably improving combined ratio.

Combined ratio measures an insurer’s underwriting profitability by comparing total expenses to earned premiums.

The formula for combined ratio is <((Incurred Losses + Operating Expenses) ÷ Earned Premiums) × 100>.

A combined ratio below 100% indicates underwriting profit. A ratio above 100% means the insurer pays out more in claims and expenses than it collects in premiums, relying on investment income to remain profitable.

Generally, anything below 100% is profitable, with ratios in the low-to-mid 90s considered strong.

Industry benchmarks vary by line of business, but combined ratios consistently above 100% are a red flag. They signal operational problems that require immediate attention.

The combined ratio has two components:

1. The loss ratio measures the total claims paid out relative to premiums.
2. The expense ratio tracks operational costs, including claims processing, underwriting, and administration.

Both components can be improved, but require significantly different strategies.

Improving loss ratios requires better risk selection, fraud detection, and claims management.

And improving expense ratios requires more operational efficiency and better processes.

Insurance carriers facing poor combined ratios often respond by raising premiums. But this approach is increasingly unsustainable.

According to the Insurance Information Institute, aggregate net premium growth across all lines reached just 5.9% recently, the lowest in over a decade. General liability in particular is struggling, experiencing its highest quarterly direct incurred loss ratio in at least 25 years.

We can attribute the slower premium growth to several factors, but a few stand out.

Unlike most other industries, insurers can’t freely change their prices. The cost of premiums, when premiums can be increased, and when changes come into effect, are heavily regulated.

Deloitte research shows that customer churn is increasing for many insurers due to inflation-driven policy-rate hikes. Rising unemployment poses an additional threat.

In lines of business where multiple carriers compete aggressively, premium increases drive customers to lower-cost alternatives. Once you lose your policyholders, it’s difficult to get them back.

The clear alternative is reducing the numerator in the combined-ratio formula. AI for claims processing helps cut both incurred losses and operating expenses through more efficient, accurate claims operations.

Fraud is the largest preventable component of incurred losses. The Coalition Against Insurance Fraud estimates fraud costs at least $308.6 billion in losses each year.

Here are three ways AI can reduce fraud losses:

Claims leakage often stems from missed details—sometimes buried deep, sometimes just poorly documented. As case workloads grow larger and more complex, claims teams need more support to accurately process documents.

Claims processing AI captures details from both image and text, whether handwritten notes or digital charts.

OwlVision extracts data with industry-best precision via a custom-curated LLM in document-processing technology that outperforms Optical Character Recognition (OCR), ensuring your team has key information to reduce claims losses or prevent overpayments.

Investigators tend to miss fraudulent claims when they overlook contradictions across multiple documents, such as:

• A claimant’s statement conflicts with medical findings;
• Employment records don’t align with disability claims; or
• Treatment patterns seem inconsistent with documented injuries.

AI fraud detection for insurance automatically flags claims that warrant investigation well before handlers authorize payments. OwlSignal unearths hard-to-find data for a layer of systematic analysis that ensures consistent scrutiny regardless of adjuster workload or experience.

According to Deloitte research, insurers that integrate AI and advanced analytics will save 20–40% on fraud costs. These savings directly lower your loss ratio and improve the insurance combined ratio.

In soft fraud where a legitimate claim is exaggerated, claimants omit disqualifying information ranging from undisclosed income sources & assets to physical activities inconsistent with claimed disabilities.

Information from external sources is critical to identifying omitted details. OwlEnrich accesses a vast network of public & proprietary sources to enrich claims data, offering automatic, real-time data retrieval that can capture:

• Public personal information
• Relations, affiliates, and employers
• Online posts, mentions, and social activities
• Legal and financial records
• Relevant news and media information
• Assets, from businesses to websites

External intelligence gives your investigators a major advantage, surfacing discrepancies that can't be found in internal documents alone, helping insurers to optimize their combined ratio.

Your expense ratio is the second component of the combined ratio. AI reduces operational expenses by increasing speed, accuracy, and employee satisfaction.

Manual document review consumes significant adjuster time, creating bottlenecks that slow the entire claims process.

Insurance document automation handles the work automatically, freeing adjusters to focus on judgment-based tasks. When documents are processed instantly rather than over days, the cost per claim is lower as claims move through workflows faster.

Additionally, when adjusters, investigators, and legal teams feel engaged in more meaningful work, it improves team morale and job satisfaction.

Conversational research with OwlAssist helps investigators find what they're looking for significantly faster. For example, instead of reading through dense medical histories to find treatment dates, investigators ask questions and receive answers with exact source references.

Reducing investigation time frees up resources, lowering your expense ratio.

Manual processes are naturally inconsistent. Different adjusters apply different standards and predictive AI tools vary in how they analyze risks—we can never be sure because of their “black box” nature.

Deterministic AI ensures every claim receives the same rigorous review and every analysis has an explanation. The combined ratio metric improves as quality becomes consistent.

Many AI solutions for insurance claims promise increased efficiency, but generic platforms come with compliance & bias risks.

Deterministic AI provides insights based on predefined rules embedded in the LLM itself, avoiding the dangers of predictive risk scores or analyses based on historical data or guesswork.

Regulations like SB26-189 in Colorado require insurers using AI to provide transparency and disclosure about decisions to avoid discrimination and bias when flagging risks.

Predictive AI creates compliance exposure since it uses existing datasets and historical patterns to make decisions that can reflect biases or make mistakes.

Ethical AI in insurance prevents these risks through deterministic AI. It evaluates each claim individually, based on documented facts. Every insight traces back to specific documents.

These measures ensure your claims operations remain completely transparent and compliant.

Technology structures information and presents insights. Humans must make the final determinations based on judgment and regulatory knowledge.

This division of labor protects your organization by ensuring decisions remain defensible. Human feedback improves AI performance, and human oversight helps you adapt during changes.

Owl enables claims adjusters & investigators to review every rationale, trace sources, configure & score insights, and make final determinations, empowering human-centric claims decisions.

Decision rationale with explainable AI for insurance is critical for improving combined ratio sustainably. With explainable AI, claims teams can make informed decisions from AI insights, thanks to the AI explaining itself.

Short-term expense reductions that create compliance violations are never worth the risk. They ultimately worsen financial performance, leading to penalties, litigation costs, and reputational damage.

Insurance leaders need to be proactive about their combined ratio. Premium growth has consistently slowed down, and unstable economic conditions can suddenly threaten revenue.

Insurance AI solutions offer an opportunity to reduce incurred losses and operational expenses, improving combined ratio over time.

But carriers should be diligent when adopting AI. Organizations need a comprehensive approach that combines powerful AI, explainable outputs, and human expertise.

Claims Intelligence explains this approach through its three principles of AI:

• Accountable AI: Explainable systems with audit trails that prevent compliance violations.
• Effective AI: Accurate, fast insights that reduce both loss ratio and expense ratio.
• Ethical AI: Fairer outcomes based on documented facts rather than predictions.

Industry leaders who achieve a good combined ratio don’t implement AI tools without a plan. They choose tools that empower human expertise, rather than replace it.

Schedule a briefing to see how the Claims Intelligence toolkit improves combined ratio accountably, effectively, and ethically.