The Cost of Unused Claims Data

Despite substantial investments in data and analytics, the insurance industry continues to overlook actionable insights within its own claim files.

‍

Ninety-seven percent of claims data, including documents, medical records, police reports, witness statements, demand letters, adjuster notes, and correspondence, remains largely unanalyzed. Although retained for compliance, this data is not actively reviewed or connected, effectively becoming digital sediment.

‍

This figure is frequently cited in industry research and acknowledged by claims executives. However, this recognition often overlooks the true impact on financial costs, outcomes, and competitive position.

‍

Key Takeaways

‍

• Many sources report that up to 97% of claims data remains unanalyzed and unused by carriers, primarily due to the overwhelming volume and complexity of claim files.
  ‍
• Insufficient use of available data results in financial inefficiencies, inaccurate payouts, and missed opportunities for fraud detection, risk assessment, and operational improvement.
  ‍
• AI-powered document intelligence and structured data analysis can bridge this gap, allowing carriers to make informed decisions, reduce losses, and remain competitive as the industry evolves.
  ‍

Practical Implications

‍

When reviewing a bodily injury claim, adjusters work with only a subset of available information. They typically review the most recent medical records, the demand letter, the police report, and sometimes the claimant’s FNOL statement.

‍

Due to time constraints, adjusters rarely cross-reference every document in a claim file. They typically do not systematically compare the mechanism of injury in the police report with emergency room records, treatment trajectories, billing patterns, claimant statements, or prior claims history.

‍

Each cross-reference is a data point that may reveal a consistency that strengthens the claim’s validity, an inconsistency that weakens it, a pattern suggesting fraud, a gap requiring investigation, or a fact that alters the liability assessment.

‍

A 1,000-page claim file contains thousands of potential cross-references. Adjusters managing typical caseloads may identify only the most apparent ones. The remainder accounts for 97% of the unused data.

‍

The cost of not knowing

‍

Consider three scenarios that occur thousands of times annually across the industry.

‍

Scenario one: The missed pre-existing condition

‍

A claimant presents with a lumbar disc herniation allegedly caused by a motor vehicle accident. The demand letter includes medical records from the treating physician and the MRI facility. However, it omits a note from a primary care visit eleven months prior, which mentions chronic lower back pain and a referral to an orthopedist. This note appears on page 743 of the complete medical records, within a 200-page section of routine primary care progress notes.

‍

If the adjuster locates this note, the claim value drops substantially. The herniation may have developed before the accident, weakening the demand and resulting in a more accurate settlement.

‍

If the adjuster does not find it, the carrier pays full value for an injury that was partially pre-existing. Industry estimates indicate that pre-existing conditions are relevant in 20 to 30 percent of bodily injury claims, leading to significant financial leakage.

‍

Scenario two: The treatment pattern anomaly

‍

A claimant undergoes twelve weeks of physical therapy for a soft tissue injury from a minor rear-end collision. Billing reflects three visits per week at an out-of-network facility with premium rates. Treatment notes, located mid-file, show minimal objective improvement from session to session. The adjuster reviews the billing total and negotiates accordingly, but lacks time to analyze the treatment-to-outcome ratio across all physical therapy notes.

‍

An AI system that reviews every note and maps the treatment trajectory would automatically flag this case: high-frequency treatment with minimal documented improvement may indicate overtreatment. This insight changes the negotiation and could save $8,000 on a single claim. Across 5,000 bodily injury claims annually, the aggregate impact is substantial.

‍

Scenario three: The litigation signal nobody caught

‍

A claim is filed in a jurisdiction known for plaintiff-friendly jury pools. The claimant retains an attorney from a firm with a history of large verdicts in similar cases. The injury type generates significant emotional sympathy. Each of these facts is present in the claim file or external data sources.

‍

Individually, none of these factors triggers an alert. Together, they form a risk profile that should direct the claim to a senior adjuster or litigation specialist immediately. However, connecting these data points requires aggregating information across the claim file, attorney databases, and jurisdictional data—a comprehensive analysis that often occurs only in retrospect, rather than in real time when prevention is possible.

‍

The Challenge of Data Volume

‍

The persistence of the ninety-seven percent figure is due to the growing volume of data. Claims files are increasing in size, medical records are more detailed, and demand packages are more extensive. Digital communication generates additional documentation, while new data sources such as telematics, social media, and wearable health data further complicate matters.

‍

Human capacity to read and synthesize information has not increased. In fact, it may be declining as experienced professionals retire and are replaced by newer adjusters with less institutional knowledge.

‍

As a result, the gap between available and utilized data is widening. The ninety-seven percent figure may have been ninety percent fifteen years ago and could reach ninety-nine percent in five years if current trends continue.

‍

Meanwhile, the plaintiff’s bar is investing aggressively in data and technology. Litigation funding firms support advanced data analytics for plaintiffs. Plaintiff attorneys use AI-powered demand letter tools, settlement value calculators, and medical records analysis platforms. The information asymmetry that once favored carriers is diminishing.

‍

Leveraging Claims Data Effectively

‍

Technology to close this gap is available today. It has already been deployed with some carriers and is producing measurable results.

‍

The approach is multi-layered. First, document intelligence: AI reviews every document in the claim file, regardless of format, and extracts structured data, including injuries, treatments, dates, providers, liability factors, and coverage terms. This process transforms ninety-seven percent of unread content into searchable, analyzable information.

‍

Second, knowledge reconstruction: the extracted data is assembled into a structured representation of the claim, forming a knowledge graph that connects injuries, treatments, providers, timelines, liability facts, and coverage terms. This is not merely a summary, but a model of the claim that can be queried.

‍

Third, reasoning: AI agents trained on insurance-specific logic analyze the knowledge graph to surface insights. Pre-existing conditions are automatically identified and flagged, treatment patterns are benchmarked against expected protocols, liability is assessed from multiple perspectives, and fraud indicators are detected through pattern recognition across the entire evidence base.

‍

The adjuster receives an integrated view of the claim. They can ask questions and receive answers supported by specific evidence from relevant documents, and can verify any finding directly at the source.

‍

The ninety-seven percent figure drops significantly. While some judgment calls will always require human interpretation, the factual foundation becomes comprehensive for the first time.

‍

Financial Impact

‍

U.S. property and casualty insurers recorded $558.8 billion in net losses incurred in 2024, putting enormous pressure on claim accuracy and file-level decision-making.

‍

At the same time, one industry source estimates claims data is 97% unstructured and warns that insurers may be making key decisions based on only about 3% of available information if that data is not effectively mined.

‍

CCC reports that the average third-party bodily injury claim payout reached $27,373 in 2024, up 38% since Q2 2020.

‍

In that environment, unused claims data becomes a real financial cost, manifesting as leakage, slower resolution, and weaker reserve and settlement decisions.

‍

Key Consideration

‍

Every carrier has data, but few use it effectively. Proven technology exists and delivers returns that far exceed the investment.

‍

The remaining challenge is organizational, not technological. When will awareness translate into decisive action?

‍

amaise transforms claims data into structured, actionable intelligence, surfacing the facts adjusters need and the patterns that drive better outcomes. Learn more at amaise.com.

‍

Frequently Asked Questions

‍

Why is it so difficult for insurance carriers to analyze all available claims data?

‍

The sheer volume and unstructured nature of claims documents make comprehensive data analysis challenging. Manual review is time-consuming and error-prone, and many legacy systems cannot handle large-scale data extraction or analysis.

‍

What are some examples of unstructured data in insurance claims?

‍

Unstructured data includes handwritten notes, emails, scanned documents, adjuster narratives, and third-party reports. These are often difficult to process with conventional analytics tools and are easily overlooked.

‍

How can insurance carriers begin to address this data utilization gap?

‍

Carriers can begin by investing in advanced technologies such as AI-powered document intelligence platforms, training staff to interpret data-driven insights, and integrating structured data protocols into claims management systems.

‍