Almost 90 percent of commercial fleets use telematics systems. Yet commercial auto insurance has posted combined ratios above 100 in 12 of the last 13 years.

The paradox deepens when you examine the data. While 88 percent of fleets have telematics programs, only 64 percent of carriers use that data in underwriting decisions. Meanwhile, 70 percent of fleet managers report they don’t share their telematics data with insurers, and 79 percent said they have not been asked.

This isn’t a technology problem. The hardware works. The data exists. The fleets are willing partners. But many carriers either don’t know how to process raw telematics data or their IT systems are not capable of processing raw telematics service provider data.

The failure extends beyond underwriting. While 88 percent of fleets have telematics capturing driving behavior, claims departments routinely process major losses without accessing available data. Risk management sends unread safety alerts. Underwriting prices renewals without loss history integration. Each department has data. None coordinate effectively to prevent losses or defend claims.

The challenge is operational. After two decades of watching telematics integration projects fail—first at Trimble, then in subsequent engagements with the insurance industry—certain patterns emerge with mathematical consistency. The same five failure modes appear regardless of carrier size, vendor selection or technical architecture.

The Five Failure Modes

Understanding why integrations fail matters more than understanding what technology to buy. The technology has been adequate for 15 years. The operational challenges remain unsolved.

Failure Mode One: The Legacy Architecture Trap

Integrations work perfectly in test environments, then collapse in production. The pattern is consistent: proof of concept succeeds, stakeholders approve, implementation begins, and the system fails under actual operational load.

The root cause is architectural, not technical. Insurance core systems—many running on mainframes from the 1970s and 1980s—were designed for batch processing, not real-time data streams. Telematics generates continuous data flows that these systems cannot handle without fundamental redesign.

TSB Bank’s 2018 core system migration provides the most thoroughly documented case. The bank attempted to migrate 1.9 million customers from a legacy mainframe to a new platform. The technical work was sound. The testing was comprehensive. The migration failed catastrophically, leaving customers locked out of accounts for weeks.

The final cost reportedly exceeded £1 billion ($1.3 billion).

Insurance faces identical challenges. Research on French motor insurers found that attempts to integrate telematics data from 2008 to 2021 repeatedly failed due to core system limitations. The carriers had the data. They understood its value. Their systems couldn’t process it at scale.

The statistics are damning. According to Clearwater Analytics, 74 percent of insurers still rely on legacy technology. Other sources reveal that the majority of IT budgets are consumed by maintaining existing systems. (See, for example, “Perspective on Modernizing Legacy Systems, PwC Hong Kong, 2024, which states that maintaining legacy systems takes up 70 percent of insurance IT budgets.) Still others find that core system modernization projects of financial institutions carry failure rates exceeding 70 percent.

This pattern extends far beyond insurance and financial institutions. During Web 1.0, I worked at Infoseek when we partnered with New Century Network—a consortium of nine leading newspaper companies including the Washington Post and The New York Times. These companies didn’t ignore the Internet; they invested early, creating websites and committing millions to digital initiatives. Yet more than 25 years later, only one successfully transformed to a digital subscription model. While multiple factors contributed to their struggles, the legacy architecture trap—print-era systems unable to handle digital-era requirements—was a critical failure mode.

Failure Mode Two: The Data Quality Mismatch

Early telematics systems were built for fleet operations, not insurance underwriting. While the necessary telematics data exists, carriers still face failure mode one and other operational challenges.

The mismatch isn’t obvious until production deployment.

The French motor insurance study documented this precisely. Insurers integrated telematics scores from fleet management systems, expecting correlation with loss ratios. The scores didn’t correlate with losses. The data measured the wrong variables.

Root Insurance’s experience seems to validate the pattern at scale in personal auto. The company built its entire business model on telematics-based pricing, achieving widespread distribution and customer adoption. By 2021, Root posted a 96.5 loss ratio—the data wasn’t predicting losses.

Academic research confirms the problem. A University of Waterloo study analyzing 28 million trips found only speeding behavior showed significant correlation with claims. Most telematics variables had no predictive value. Even with massive datasets, prediction accuracy rarely exceeds 50 percent—barely better than random chance.

Related article: “Speeding Is Riskiest of all Bad Driving Behaviors: Research“

The brutal reality: 88 percent of fleets have telematics programs, but only 64 percent of that data gets used because the data doesn’t match underwriting requirements.

The consequences extend to litigation and loss prevention. Plaintiff attorneys routinely ask what operational or safety factors could have prevented crashes—yet fleets with comprehensive telematics data often cannot demonstrate they acted on safety warnings.

The disconnect between data collection and operational response transforms potential evidence of diligence into proof of negligence. Carriers collect data they cannot use.

Failure Mode Three: The Workflow Adoption Problem

Systems that work technically often sit unused operationally. The integration functions. The data flows. Underwriters ignore it.

The system requires underwriters to change established workflows. They resist or work around the new tools.

If operational leadership doesn’t use the tools, no one will. Behavior cascades through organizations. When a vice president continues using spreadsheets instead of the new system, the entire team follows. Technology purchases don’t change behavior without leadership modeling.

The data on this failure mode is extensive. Accenture surveyed insurers in 2008 and again in 2021, measuring whether new technology reduced workload. In both years, only 35 percent said technology decreased their workload. Sixty-four percent reported workload stayed the same or increased.

Related articles: The Road to 2032: Big Changes Ahead for Commercial Underwriters; Where Carriers Are Spending: The Investments Driving Underwriting Change

Fleet risk management reveals the same pattern. The 2025 SambaSafety Telematics Report found that 66 percent of fleets identify “interpreting or acting on the data” as their primary challenge. Driver coaching programs exist. Real-time alerts exist. The technology captures aggressive driving, speeding and harsh braking. Fleet managers receive reports they cannot translate into behavior change. The systems generate data overload without actionable insights.

Thirteen years of technology investment. Zero improvement in adoption.

The reasons are clear: technology changes workflow, but workflow changes require organizational change management that most carriers never consider until after deployment fails.

AI and machine learning integrations follow identical patterns. Carriers invest in predictive models, then discover underwriters don’t trust the outputs or can’t incorporate them into existing workflows. The technology works. The adoption fails.

Consider the scaling data: BCG research reveals a 42 percent level of AI adoption among insurers but only 7 percent have brought their AI systems to scale beyond pilots. The gap between technical success and operational adoption is enormous.

Failure Mode Four: The Timeline Mismatch

Major integration projects require 25 to 35 months from contract signing to production deployment. During that timeline, three things happen with predictable consistency: the vendor gets acquired, the internal champion leaves, or organizational priorities shift.

The mathematics are brutal. Core system implementations take 25 to 35 months on average. Chief information officer tenure averages between 3.9 years and 4.7 years in the insurance industry. The implementation timeline isn’t always shorter than the executive’s expected tenure. In many situations, if the CIO who championed the project leaves, the new CIO often cancels it to pursue their own initiatives.

Vendor consolidation accelerates the problem. Based on operational experience during a period when major insurance technology vendors underwent multiple ownership changes, the disconnect between carrier integration needs and platform vendor capabilities was substantial. Telematics integration, despite being a growing carrier priority, often fell outside core vendor road maps during ownership transitions.

Vertafore provides a clear example: The company changed ownership three times in 14 years, with sale prices escalating from $1.4 billion to $2.7 billion to $5.35 billion. Each ownership change brought new strategic priorities and road map revisions.

Prudential’s $2.35 billion acquisition of Assurance IQ ended with the platform shut down after five years. Customers and partners received no warning. Projects depending on the platform failed overnight.

Failure Mode Five: The Department Silo Catastrophe

The IT-underwriting trust gap represents only one dimension of a broader organizational failure. Successful telematics integration requires coordination across underwriting, claims and risk management. Most carriers run these departments as independent operations with incompatible systems and conflicting priorities.

The statistics are stark. While 88 percent of commercial fleets have telematics and 21 of the top 25 insurers have implemented telematics programs, departments within the same carrier often use different vendors and cannot share data. Underwriting collects driving behavior for pricing. Claims departments have their own telematics data for investigation. Risk management runs separate driver coaching programs. None of these systems talk to each other.

The integration failures compound. An underwriter prices a risk based on fleet safety scores. Six months later, the same fleet has a major accident. The claims department has telematics data showing the driver was speeding—but that information never reaches underwriting for the renewal. Risk management has been sending unread safety alerts to the fleet manager for months. Each department measured success by its own metrics. All three declared their programs successful. The carrier posted an underwriting loss.

When plaintiff attorneys ask why available safety data wasn’t acted upon, the answer reveals the silo problem: the data existed in three different departments, none of which coordinated with the others. Technology integration succeeded in each department. Organizational integration never happened.

Why Conventional Wisdom Fails

The standard recommendations for fixing integration problems don’t work. They sound logical. They follow best practices from other industries. They consistently fail in insurance operations.

“Modernize core systems first” is the most common advice. The logic seems sound: fix the foundation before adding new capabilities. The reality is that core system modernization has failure rates that exceed 70 percent and requires five to seven years to complete. Waiting for modernization means waiting forever. The market moves faster than core system timelines.

“Use more data” assumes volume solves quality problems. Carriers already have more data than they can analyze. The University of Waterloo study examined 28 million trips and found most variables had no predictive value. More data amplifies noise unless you know which signals matter.

“Build comprehensive solutions” creates projects too large to complete. Scope expands. Timelines extend. Stakeholder alignment becomes impossible. Only 7 percent of comprehensive initiatives scale beyond pilots.

“Force process change” guarantees adoption failure. Without culture change, IT spending fails to deliver a return on investment. Technology purchases don’t solve problems—they require business process re-engineering that most carriers never attempt. Leaders mandate new tools, then wonder why usage metrics stay at zero.

The pattern transcends technology categories. Carriers invest in predictive models that underwriters don’t trust. They deploy automation that requires more manual work than the old process. They build data lakes that nobody queries.

The common thread: technology implementations that ignore workflow realities fail regardless of technical quality.

The Questions Successful Integrations Answer

Evaluating integration approaches requires asking different questions than most carriers consider. These aren’t implementation questions. They’re filtering questions that expose whether a proposed solution will work in your operational environment.

Does the system adapt to current workflow or force workflow change? This question predicts adoption rates more accurately than any technical specification. Systems requiring significant workflow changes fail unless leadership commits to multi-year change management programs. Most carriers lack that commitment.

Does the data match underwriting criteria or provide generic fleet metrics? Fleet operational data serves fleet managers. Insurance underwriting data serves underwriters. These are different datasets. Verify the data captures variables that correlate with your loss experience, not variables that seemed useful to someone optimizing logistics.

Can you prove value in 90 days, or does implementation require multi-year timelines? Long timelines guarantee failure through the mechanisms described in Failure Mode 4. Champions leave. Vendors get acquired. Priorities shift. The project dies. Approaches that can’t demonstrate value quickly won’t demonstrate value ever.

Do IT and underwriting agree on success metrics before signing the contract? If both departments don’t define success identically—and commit to that definition in writing—the project will succeed technically and fail operationally. Misaligned metrics guarantee someone declares victory while someone else declares defeat.

What happens when the vendor gets acquired or the platform goes down? Amazon Web Services experienced a significant outage recently. Some customers continued operating. Many others went dark despite multi-region architecture because some AWS services depend on a single region. Even well-architected disaster recovery plans failed.

Have a documented answer for “then what?” or accept that your integration will eventually fail without warning.

These questions don’t prescribe solutions. They expose whether proposed solutions will survive contact with your operational reality.

The Integration Imperative

Commercial auto insurance posted combined ratios above 100 in 12 of the last 13 years. Median nuclear verdicts for auto have exceeded $20 million for a number of years.

The technology to fix this exists. It has existed for 15 years. Telematics can identify high-risk operators. Data integration can improve underwriting accuracy. Risk management can change driver behavior. The barriers aren’t technological. They’re operational.

This creates a competitive opportunity. Carriers that solve the integration challenge will dominate. Those that don’t will exit the segment or sign white label agreements with carriers who have solved this. The margin between success and failure is execution, not innovation.

Two decades of failures provide the road map. We know why integrations fail. We know which patterns repeat. We know which approaches consistently produce the same failures.

The question isn’t whether telematics integration matters. The question is whether carriers can learn from documented failure patterns before market conditions force them out of commercial auto entirely.

Understanding why everyone else fails is the prerequisite for not failing yourself.

AI Disclosure: Research for this article utilized Perplexity AI to discover and verify publicly available data sources and citations. All analysis, interpretation, conclusions and writing are original work by the author based on 20 years of operational experience leading telematics data integration programs at @Road/Trimble and working with major commercial auto insurance carriers.