Samsara Report Finds 10% of Drivers Account for Nearly Half of Crashes
Samsara has released new research highlighting how a relatively small group of high-risk drivers can account for a disproportionately large share of crashes across commercial fleets. The company’s new Compounding Risk Report, based on its patent-pending Risk Model, finds that the highest-risk 10% of drivers are associated with 47% of crashes.
The findings point to a potential shift in how fleet operators approach driver safety. Rather than treating every safety event independently, the research suggests that fleet managers can achieve greater impact by identifying recurring combinations of risky behaviors and concentrating coaching resources on the drivers who demonstrate the strongest patterns of elevated risk.
Samsara, a provider of the Connected Operations Platform, developed its Risk Model by examining approximately 50 different factors related to driving behavior, exposure, operating context and driver development. The objective is not to determine whether a particular driver will crash on a specific day. Instead, the system ranks drivers according to their relative risk, allowing safety teams to determine where intervention and coaching may be most valuable.
Risk Is Concentrated Among a Small Group of Drivers
One of the most significant conclusions from the report is the degree to which crash risk is concentrated within a relatively small segment of the driver population.
According to Samsara’s analysis, the top 10% of risk-ranked drivers account for 47% of crashes. When the group is expanded to the top 30% of drivers, the share rises to 76% of crashes.
The results indicate that fleet safety teams may not need to distribute their most intensive coaching efforts equally across an entire workforce. Instead, identifying the drivers showing the strongest combination of risk indicators could allow managers to use their available time more efficiently.
For organizations operating hundreds or thousands of commercial vehicles, this distinction can be particularly important. Safety managers often have limited time for one-on-one coaching, reviewing video events and following up on recurring driving behaviors. A system capable of identifying where risk is most concentrated can help them prioritize those activities.
Arpan Podduturi, Head of Safety Product at Samsara, said focusing intensive coaching on the highest-risk 10% could give managers an opportunity to reach a group associated with nearly half of crashes.
The approach also extends beyond formal manager-led coaching. Once safer driving practices are established, a strong safety culture and self-coaching can help reinforce those behaviors among the wider workforce.
Multiple Risk Factors Can Compound Each Other
Another important finding from the research is that risky driving behaviors should not necessarily be viewed in isolation.
A driver who demonstrates one potentially dangerous behavior may present a different level of risk from a driver who repeatedly demonstrates several behaviors at the same time. Samsara’s research indicates that combinations of behaviors can provide a stronger signal of elevated risk than individual behaviors considered separately.
Mobile phone use provides one example. Drivers exhibiting mobile use alone were 2.7 times more likely than the overall driver population to be classified in the highest-risk tier.
When harsh braking was added to mobile use, the risk concentration increased substantially, reaching 4.5 times the overall population level.
The combination became even stronger when mobile use, distraction and harsh braking occurred together. In that situation, the associated risk concentration reached 5.4 times that of the overall driver population.
These findings illustrate the concept behind the report’s title: risk can compound when several behaviors occur together.
For fleet safety teams, this could mean that simply counting individual safety events may not provide a complete picture. A driver who receives separate alerts for mobile phone use, distraction and harsh braking may require a different intervention strategy from someone who experiences an isolated event.
Persistent Patterns May Offer Earlier Warning Signs
The report also examines whether risk-related patterns persist over time.
Samsara compared drivers who were subsequently involved in a crash with drivers who were not and found that its model prioritized the crash-involved driver approximately three out of four times.
This result remained consistent across both next-day and seven-day evaluation periods.
The finding is important because it suggests that safety managers may have an opportunity to intervene before a crash occurs rather than relying solely on post-incident investigations.
Traditional fleet safety programs often respond after an accident, reviewing what happened and determining how similar incidents can be prevented. While post-crash analysis remains important, identifying persistent behavioral patterns beforehand could allow companies to take preventive action.
The Risk Model is therefore designed to help managers recognize patterns that may otherwise be difficult to identify when looking at individual safety events separately.
Aggressive Driving, Distraction and Speeding Stand Out
Not all risk factors identified by the model have the same implications for coaching.
Samsara’s separate causal analysis identified aggressive driving, distracted driving and speeding-related patterns as the most consistent and credible behavioral contributors to crash risk.
These behaviors are particularly relevant because they are potentially coachable. Unlike environmental conditions that drivers may have limited ability to control, behaviors such as speeding, aggressive maneuvers or distracted driving can be addressed through training, feedback and changes in driving habits.
The distinction is important for safety managers.
For example, night driving may increase exposure to certain hazards, while freezing temperatures can create additional road challenges. Urban driving can also involve greater traffic density, pedestrians, intersections and other sources of complexity.
However, these conditions are not themselves coaching targets in the same way that distracted or aggressive driving can be.
Instead, environmental and operating conditions can be viewed as factors that may amplify existing risks. A safety program can therefore use them to provide context while concentrating coaching on behaviors that drivers can change.
AI Helps Fleet Managers Prioritize Coaching
The findings from the report are connected to Samsara’s Coaching Priority feature, an AI-powered capability designed to help fleet operators identify and prioritize coaching opportunities.
Coaching Priority considers approximately 50 different risk factors and combines those signals into a single view for safety managers.
Rather than forcing managers to review every driver and every event with the same level of attention, the system is designed to highlight where intervention could potentially have the greatest effect.
The platform can provide greater visibility into the drivers, behaviors and locations associated with elevated risk. This can help safety professionals move from reactive event management toward a more proactive approach based on recurring patterns.
Tom Karnowski, Vice President of Environmental, Health, and Safety at USIC, described the system as providing clarity about where the company should concentrate its safety efforts, including specific locations, behaviors and drivers contributing to risk.
The concept is particularly relevant for large fleets, where the sheer volume of telematics and video data can make manual analysis difficult.
Moving From Individual Events to Broader Patterns
Modern commercial vehicles can generate enormous amounts of operational data. Cameras, telematics systems and connected vehicle platforms can record braking events, acceleration, speeding, phone use, distraction and other driving characteristics.
However, collecting information is only part of the challenge. Fleet operators also need to determine which signals matter most and which should trigger action.
The Compounding Risk Report emphasizes the importance of looking beyond isolated incidents.
For example, a single harsh-braking event may not necessarily indicate a persistent safety problem. But repeated harsh braking combined with distraction or mobile phone use could create a more meaningful pattern.
This approach potentially allows safety teams to distinguish between occasional events and recurring behaviors.
It can also make coaching more specific. Instead of telling a driver simply to “drive more safely,” a manager can focus on the particular behaviors contributing to the driver’s elevated risk profile.
Self-Coaching Can Extend Safety Programs
Manager-led coaching remains an important component of fleet safety, but it can be difficult for managers to provide continuous individual attention to every driver.
Samsara’s approach combines targeted managerial coaching with automated self-coaching. The idea is that managers can focus their most intensive efforts on drivers with the strongest risk signals, while technology helps reinforce safer practices across the broader workforce.
This can support a safety culture in which drivers become more aware of their own behaviors.
Over time, the goal is not simply to respond to individual violations but to encourage consistent improvements in driving habits. A combination of data, targeted feedback and driver engagement can potentially make safety programs more continuous rather than event-driven.
Risk Ranking Is Not Crash Prediction
Samsara emphasizes that its Risk Model should not be interpreted as a system that predicts whether an individual driver will crash on a particular day.
Instead, it ranks drivers according to relative risk.
That distinction is important when interpreting the report’s findings. A driver classified within a higher-risk group is not guaranteed to be involved in a crash, just as a driver classified at lower relative risk is not guaranteed to avoid one.
The model is intended as a prioritization tool.
By ranking relative risk, it can help fleet managers determine where limited coaching resources may have the greatest potential impact. Managers can then combine those rankings with their own knowledge of drivers, routes, operating conditions and company safety policies.
Research Draws on Data From Multiple Industries
The research behind the report is based on aggregated data used to train and evaluate Samsara’s Risk Model between July 1 and December 15, 2025.
The dataset includes drivers and fleets operating across multiple industries and regions, providing a broad foundation for the analysis.
Samsara also makes clear that the combined-profile findings represent statistical associations rather than direct proof of causation.
The company conducted separate causal analyses to identify behaviors that appear to make more credible contributions to crash risk. Those analyses used double machine learning, calibration and causal forests for policy ranking.
This methodological distinction is important because an association between two variables does not automatically establish that one caused the other. Samsara’s research therefore separates the statistical identification of risk patterns from its analysis of potentially causal behavioral factors.
A Potential New Direction for Fleet Safety
The findings arrive as fleet operators increasingly rely on connected vehicle technology and artificial intelligence to manage safety.
For many companies, the challenge is no longer simply collecting safety information. The larger challenge is turning that information into useful decisions.
Samsara’s research suggests that concentrating on a relatively small number of high-risk drivers could be an effective way to address that challenge.
If the top 10% of risk-ranked drivers are associated with 47% of crashes, identifying that group could allow safety teams to direct intensive resources where they may matter most. Expanding attention to the top 30% could cover a group associated with 76% of crashes.
At the same time, the research indicates that safety programs should consider combinations of behaviors rather than isolated events. Mobile use, distraction and harsh braking, for example, appear to create a substantially stronger risk signal when they occur together.
Ultimately, the report presents a data-driven approach to fleet safety in which technology helps managers identify recurring patterns, prioritize interventions and reinforce safer behavior over time.
For fleet operators, the message is straightforward: improving safety may not require treating every driver and every event identically. By identifying concentrated and compounding risk, managers may be able to focus their limited resources more strategically while building a broader culture of safer driving.
Samsara’s Compounding Risk Report provides detailed methodology, analysis and recommendations for applying these findings within fleet safety programs. The company notes that its combined-profile findings describe statistical associations and should not be interpreted as predictions of individual crash outcomes.
Source Link:https://www.businesswire.com/
