Commercial auto insurance has posted consecutive underwriting losses in the last 14 years, according to AM Best. Combined ratios have exceeded 103% from 2022 to 2024, and insurers have absorbed billions in net underwriting losses recently, despite consecutive quarters of rate increases.
The persistence of those losses, even through an extended period of rate hardening, points to a fundamental issue. Rate increases respond to losses that have already occurred. Underwriting accuracy can help prevent them. For commercial auto insurers looking to build sustainable profitability rather than chase it, the distinction matters enormously.
This post outlines what data-driven underwriting accuracy can look like in practice—and what sets insurers who are closing the gap apart from those still widening it.
Why Underwriting Profitability Depends on Data
Rate increases in the broader commercial insurance market are slowing down. WTW’s Commercial Lines Insurance Pricing Survey (CLIPS) reported an aggregate growth rate of just 2.9% for U.S. commercial insurance in the fourth quarter of 2025, compared to 5.6% in the same quarter of 2024. Commercial auto liability has continued to rise. Still, rate growth across the market may be slowing, while claims severity, influenced by social inflation and nuclear verdicts, has outpaced economic inflation for a decade. According to AM Best, the average commercial auto claim cost more than doubled over the past decade, rising at roughly 8% annually.
The gap between what insurers are charging and what claims are costing them isn’t something another round of rate increases will reliably close. Better information at the point of underwriting can help close it.
A commercial auto submission that arrives with accurate driver data, verified fleet composition, and complete loss history can be priced more accurately from day one. One that arrives with self-reported driver rosters, misclassified exposures, and missing loss runs may be priced on assumptions. And in this environment, assumptions can be expensive. According to AM Best, among the top 20 commercial auto insurers, 14 posted combined ratios exceeding 100 in 2024. The six remaining insurers may be doing something different at the front of the underwriting process.
The Case for Driver-First Underwriting Data
Loss runs tell underwriters what happened on prior policies. Driver data can tell them what’s likely to happen next. The most predictive variable in commercial auto underwriting may not be prior loss experience; it could be the behavior of the drivers operating the insured vehicles.
Many commercial auto submissions still arrive with incomplete or unverified driver rosters. Fleet operators self-report their drivers, and underwriters either accept those rosters at face value or spend time they don’t have verifying them. Neither approach produces optimal risk assessment.
Improving underwriting accuracy can start with knowing, with greater confidence, who is driving the insured vehicles—license status, violation history, and whether the record presented matches what state DMV data actually shows. Motor vehicle records (MVRs) remain among the most reliable sources of that information, but how and when they’re obtained may matter as much as whether they’re obtained at all.
Ordering a full MVR on every driver at every submission is increasingly costly. As state DMV fees continue to rise, the case for a more strategic approach grows stronger. Pre-screening first—using violation indicators and license verification to triage the driver roster before committing to full MVR spend—can allow insurers to fast-track clean risks, surface violations that warrant closer review, and help identify identity mismatches before significant spend is incurred. The goal is to spend the data budget where it produces the most underwriting value, not uniformly across every submission regardless of risk profile.
Underwriting Fraud Detection Starts at Submission
Application fraud in auto insurance has grown more sophisticated in recent years. The foreign driver’s license (FDL) evasion pattern, drivers falsely claiming FDL or unlicensed status to avoid MVR pulls and violation discovery, remains common. Generative AI tools now also enable synthetic identity creation at scale, producing fabricated driver profiles that pass initial screening.
At the fleet level, a single fraudulent driver buried in a large roster can generate losses that exceed the entire account premium. That risk can be most cost-effectively addressed at the time of submission. Confirming a valid U.S. license, verifying driver identity against state records, and uncovering hidden license history before a policy is bound has moved from best practice to baseline requirement for insurers managing commercial fleet exposures.
Identity verification at submission is one place where underwriting fraud risk can be caught earlier. Waiting until claims to discover it is a fundamentally reactive and expensive approach.
Underwriting Automation and Straight-Through Processing for Clean Risks
Not every commercial auto submission needs the same level of underwriting attention. Accounts with clean driver records, verified loss history, and exposures that fall within appetite can be evaluated and bound through automated workflows without significant manual intervention—freeing up underwriting capacity for submissions that actually require judgment.
Straight-through processing (STP) routes those clean risks through a predefined rules engine and returns a bindable quote without manual review. Underwriting resources can then be reserved for submissions with adverse loss history, complex exposures, or driver records warranting closer scrutiny. The result can be faster turnaround on straightforward accounts and better decisions on the complex ones.
The prerequisite is data quality. An automated rules engine can only be as accurate as the data it receives. Submissions with incomplete driver rosters, missing loss runs, or unverified identity information may produce incorrect automated decisions at scale. The driver verification steps above aren’t separate from STP; they’re what make it reliable.
Predictive Analytics in Underwriting Risk Assessment
Traditional commercial auto underwriting is largely retrospective. Loss runs describe what happened on prior policies. MVRs record violations that have already been documented. Safety program documentation reflects what a fleet reports about its own practices. Taken together, these inputs describe the past—not necessarily the current or forward-looking risk profile of an account.
Predictive analytics in underwriting can shift that frame. By modeling relationships between driver behavior patterns, fleet characteristics, violation history, and loss outcomes across large datasets, predictive approaches may identify risk signals that point-in-time snapshots miss entirely.
The most predictive signals may not always be the most obvious ones. A driver with a clean MVR but a pattern of hard braking and speeding captured in telematics data could present a meaningfully different risk than the MVR alone would suggest. A fleet with documented safety programs but high driver turnover may carry a different level of exposure than a stable, well-monitored fleet with a similar loss history.
Telematics data can be increasingly central to this analysis. According to SambaSafety’s 2025 Telematics Report, 87% of commercial fleets now use telematics to manage safety, suggesting behavioral data may be available for most of the accounts commercial auto insurers write. Building a more complete predictive picture can mean bringing together:
- MVR monitoring and violation history
- Telematics behavioral data like speed, braking, time-of-day, and distraction events
- Training completion records and coaching responses
- Documented fleet safety program adherence
The challenge is normalizing that data across device types and telematics service providers (TSPs). Telematics aggregation can address that directly, potentially giving underwriters a more unified view of driver behavior regardless of the fleet’s system.
Why Underwriting Risk Assessment Doesn’t Stop at Bind
One of the most persistent gaps in commercial auto underwriting may be treating driver risk as a point-in-time assessment rather than an ongoing one. An MVR pulled at renewal reflects a driver’s record at a single point in time. A reckless driving conviction or a license suspension occurring six months into a policy period may not appear until the next renewal pull, if one is ordered at all.
Commercial auto liability losses are driven by incidents, not by what underwriters knew at bind. Drivers whose risk profiles change materially during a policy period can represent exposure that annual or renewal-only monitoring may miss entirely.
Midterm monitoring—automated alerts when adverse driving activity or license changes occur on covered drivers—can help close that gap. It may allow underwriting and risk control teams to respond to emerging risk profiles by recommending driver training, adjusting terms, or initiating non-renewal processes before a high-risk driver generates a claim. The shift away from annual MVR pulls toward continuous monitoring may represent one of the highest-impact changes available to commercial auto underwriters. It is also one of the least widely implemented, which means it may remain a meaningful competitive differentiator for insurers that adopt it.
Data Quality as a Core Strategy
Each of the strategies above depends on data quality. Predictive models trained on poor data learn poor patterns. Automated rules applied to unverified submissions produce incorrect decisions. Pre-screening that misses identity fraud fast-tracks the risks it was designed to catch.
Insurers that are improving their commercial auto underwriting results treat data quality as a strategic function, not a back-office maintenance task. In practice, that means four things:
1. Validate data at submission, not after binding
Problems identified before a policy is written may cost nothing to fix. Problems discovered at claims can cost significantly more.
2. Integrate data sources systematically
Driver violation history, license status, identity verification, and telematics data can automatically flow into the underwriting workflow—rather than requiring manual assembly across disconnected systems.
3. Audit data sources regularly
Third-party data can degrade over time. State DMV feeds may vary in timeliness and completeness. Telematics device data has known quality issues at scale. Regular auditing can catch systematic data problems before they affect decisions across the book.
4. Close the loop between underwriting and claims
Some of the most valuable data for improving underwriting accuracy may be internal. Claims outcomes on prior accounts can reveal which risk signals were weighted correctly and which were missed. Insurers who incorporate that feedback into their underwriting models may continuously improve over time.
A Connect Strategy to Improve Underwriting Efficiency
Commercial auto’s persistent underwriting losses may not be primarily a rate problem or a market cycle problem. They can reflect the challenge of making accurate risk decisions with incomplete, unverified, and point-in-time information.
Insurers posting better results in this environment may be doing so by building data infrastructure that supports better decisions at every stage of the policy lifecycle—from submission through the policy period to claims. Driver identity verification, MVR pre-screening, straight-through processing for clean risks, predictive analytics, telematics integration, and continuous monitoring are each available today. Insurers whose combined ratios are moving in the right direction are often those treating them as a connected strategy rather than a collection of independent initiatives.
SambaSafety provides driver risk assessment and underwriting data solutions for commercial auto insurers. Learn how leading insurers are using data to improve underwriting accuracy and reduce loss ratios today.
-1.png)