Skip navigation

The value of seeing and acting on driver risk factors

C.R. England reports on predictive risk modeling of its drivers

It takes information, analysis and-- above all-- action to mitigate the risks truck drivers bring to the road at any given time. That’s what giant refrigerated carrier C.R. England realized and benefitted from via its implementation of a predictive-modeling solution called the “Fleet Safety Model” provided by FleetRisk Advisors.

During a recent webinar co-hosted with C.R. England, FleetRisk Advisors detailed how the computerized system uses advanced analytics to predict which drivers in a fleet are most likely to have an accident during the next 28 days. The firm customizes the safety model for each trucking company by drawing from a variety of data sources, including operations, human resources and safety.

To “identify risk signatures of drivers, vehicles, and schedules most likely to be involved in an accident,” the model “uses key week-to-week predictors such as change in length of haul and hard braking.” It also measure empty miles and time idling vs. the fleet average and monitors the days since each driver’s last vacation.

The ongoing, 28-day cycle of data analysis includes focusing on the types of accidents that need to be prevented, the most critical risk factors and the highest-risk drivers at a given time. In this way, stated stated Chris Orban, director of technical services for FleetRisk Advisors, the model can predict the future safety performance of drivers.

The upshot, according to both FleetRisk Advisors and CR England, is the fleet can intervene with drivers before they have an accident to lower the company’s risk exposure and thus reduce costs by reducing the frequency and severity of accidents.

Indeed, according to C.R. England, since instituting the Fleet Safety Model, it has attained a 32% reduction in frequency of accidents; an 81% reduction in severity of accidents; a 17% reduction in driver turnover, and a 24% increase in productivity.

But it’s not all about data-crunching. Both firms explained during the webinar what makes the model work is how the information it generates is used by the fleet’s owner driver-management personnel.

“We run 4300 tractors operated by 6200 company drivers and independent contractors,” said Dustin England, the carrier’s corporate vice president of safety/compliance. “Most of our hires are new to the industry and we put them through one of our five driving schools. We ban using predictive modeling to reduce driver risks in 2008, pioneering this technology in trucking. Our idea was to become proactive about risks rather than reactive and we adjusted our original target of cutting cost per mile to reducing the number of preventable accidents.”

England explained what the carrier does with the data generated by the predictive model: “Every 28 days we get a score on each driver and we go out and talk [either in person or by phone] with the 10% of our drivers who are predicted to be at the highest risk of an accident in the next 28 days.”

He said these conversations often reveal what is on the driver’s mind that could be distracting him from performing well. These run the gamut from skills/training issues to worries over finances or personal and family matters. From there, remediation of the problem is attempted via a “Recommended Action Plan” prepared for the individual driver and his or her circumstances, be they strictly work-related or of a more personal nature.

“We use what the model tells us about the drivers and what we learn from them to arrive at the right interventions,” England continued. “For example, a change in pay could trigger a talk with them about managing finances or an increase in inspection violations would indicate the need for more training.”

He pointed out that since starting down this path, CR England has not terminated anyone due to their 28-day risk scores as the whole idea is “to figure out the issue behind a negative score and go from there to address it.”

England added that the carrier has learned that it is more effective, when speaking to drivers about their risk situation, to “not discuss predictive modeling per se, but to make the conversations about the driver while pointing out they are not the only one being contacted in this regard.”



Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.