Data analysis proves beneficial in driver recruiting, retention

user-gravatar Headshot
Updated Nov 25, 2015
Factors such as driver perceptions and motivational behaviors can be far more complex to manage than home time, miles or model year of equipment.Factors such as driver perceptions and motivational behaviors can be far more complex to manage than home time, miles or model year of equipment.

When the driver turnover rate at Milan Supply Chain Solutions surpassed 100 percent this year, management for the Jackson, Tenn.-based fleet decided to analyze its data to understand why.

In July, the data showed that 75 percent of terminated drivers were not married, 17 percent were rehires, and 88 percent had gaps in their employment of six months on average within the past three years.

David Dallas, senior vice president of the 350-truck carrier, shared these findings during a panel discussion on driver recruiting and retention at last month’s McLeod Software user conference in Birmingham, Ala.

Milan Supply Chain Solutions changed its on boarding practices after this analysis in an effort to get turnover down to 40 percent or less, he said. Knowing that engaging the spouse is important to retention, the company now spends more time explaining health insurance and other benefits of interest to drivers and their families.

As for rehires, Milan now funnels them through safety, operations and human resources to get additional screening before making a final decision.

While the company also has become more selective of drivers with employment gaps – particularly those who come through its driving school – enforcing this policy has been difficult.

“We all have the emotion of ‘how we are trying to help people?’ ” he says. When this policy was relaxed, “the statistics showed us again that they didn’t last.”

The trucking industry is expected to be short nearly 50,000 drivers by the end of the year, according to the American Trucking Associations. Besides offering better pay packages, benefits and incentives, fleets are looking harder at their own data to uncover the reasons why drivers are leaving and to prevent turnover, where possible, by taking action on early warning signs.

Looking for trends

Other fleet executives who spoke on the panel during the McLeod Software conference shared some insights into what they are seeing in their data.

Interstate Distributor Co., a dry van and refrigerated truckload carrier based in Tacoma, Wash., has noticed that drivers who are referred stick around longer.Interstate Distributor Co., a dry van and refrigerated truckload carrier based in Tacoma, Wash., has noticed that drivers who are referred stick around longer.

Interstate Distributor Co. (CCJ Top 250, No. 79), a dry van and refrigerated truckload carrier based in Tacoma, Wash., has noticed that drivers who are referred stick around longer, said Paul Simmons, chief operating officer.

Decker Truck Line (No. 135), a 700-truck refrigerated and flatbed carrier based in Ft. Dodge, Iowa, sees higher turnover for drivers recruited from training schools. New drivers also have more accidents, said Jennifer Brim, director of fleet management.

Decker had shifted away from recruiting drivers from schools but reverted back due to the labor situation. The company now requires that drivers have at least one year of experience and offers training to bridge the gap.

Direct feedback

Driver rewards and loyalty programs are used by 45 percent of fleets, according to a Sept. 3-4 survey by CCJ. The programs typically work by rewarding points for attaining periodic goals and milestones in categories such as safety, compliance, fuel savings and tenure. Drivers redeem their points towards noncash items.

About 10 percent of fleets that have a rewards program use a third party to administer it, the study found.

“We have found that a well-structured rewards program with obtainable rewards for a driver can truly affect turnover,” says John Elliott, chief executive officer of Load One. The Taylor, Mich.-based ground expedite carrier has been using a rewards platform from Stay Metrics for three years. “It helps to make companies with good culture even better,” Elliott says.

Fleets that run their own rewards program gave their effectiveness a 7.25 average score on a scale of 10, with 10 being the most effective. Fleets that have a third party manage their program give their program a score of 7.8.

The study, which had 109 respondents, also found that rewards programs are used most widely by fleets with between 50 and 500 trucks.

Tim Hindes, Stay Metrics CEO, says the company’s own research shows that drivers who are engaged in a rewards program – those who use its rewards site at least three times per month – have a 24 percent lower turnover rate.

Workhound created a cloud-based management tool and a mobile app that aggregates feedback data from drivers and shares insights that companies can use to better manage their relationships with them.

Tucker Robeson led the startup of CDL Helpers, a company that created its own cloud-based Fleet Relationship Management application. The app is designed to keep a rolling log of driver interactions to gather useful information and ensure those encounters follow a consistent – and unique – process for each carrier.

Over time, the app is able to benchmark conflicts entered in the system – such as a driver being frustrated with a dispatcher – against the long-term success rates of drivers. This benchmark serves as a useful predictor for how many hours the company has to neutralize the threat before it escalates and leads to turnover, Robeson says.

Predictive intelligence

In 2005, Lafayette, La.-based Dupre Logistics tapped FleetRisk Advisors to create predictive models to analyze complex data sets and identify drivers who were the most likely to quit.In 2005, Lafayette, La.-based Dupre Logistics tapped FleetRisk Advisors to create predictive models to analyze complex data sets and identify drivers who were the most likely to quit.

Ten years ago, Lafayette, La.-based Dupre Logistics tapped a startup company named FleetRisk Advisors to create predictive models to analyze complex data sets and identify drivers who were the most likely to be in accidents, file workers compensation claims or quit within the next couple of weeks. More fleets signed up, and the models got even better, with more experience and data fueling the engines.

Advanced statistical methods are able to find the patterns in operational data that are predictive of future events. The models also help determine the best countermeasures to apply to mitigate the risk.

Such countermeasures may be the suggested topics of conversations for fleet managers to have with drivers about personal, professional or financial issues they are having. This insight is derived from patterns in data that otherwise might go unnoticed.

To date, experience has shown that using predictive information is more effective in preventing driver turnover than for targeting accidents, says Dean Croke, a FleetRisk Advisors founder. The reasons drivers quit usually are the same, but circumstances that lead to accidents are more complex and less repeatable.

“Drivers are very predictable – they experience the same frustrations,” says Croke, who is now vice president of Omnitracs Analytics, FleetRisk Advisors’ new name following the purchase of Omnitracs and its FleetRisk subsidiary by Vista Equity Partners in 2013.

Croke says each client has unique predictors, but in most cases, the early signs of turnover are things you might expect would cause drivers to be frustrated.

Common predictors include fluctuations in pay and mileage, detention time at docks and denied requests for home time. A less-obvious predictor is the geographical state where drivers hold their commercial driver’s license; over-the-road drivers who live in states with low freight volumes may have more difficulty getting home. “It’s obvious from the start that the driver is going to quit,” Croke says.

Decker Truck Line, a 700-truck refrigerated and flatbed carrier based in Ft. Dodge, Iowa, sees higher turnover for drivers recruited from training schools.Decker Truck Line, a 700-truck refrigerated and flatbed carrier based in Ft. Dodge, Iowa, sees higher turnover for drivers recruited from training schools.

On the flip side, these and other data patterns also help explain why drivers stay. Drivers who stay longer have markedly different traits than drivers who quit under the same circumstances. By identifying these traits in their profile data, it is possible to know from the start the probability that a job applicant will leave early or stick around, Croke says.

The challenge of using predictive intelligence on the front end of the employment lifecycle, he says, is the additional hurdles it can create to keep drivers moving through the recruiting pipeline. Fleets often need to hire all of the qualified drivers they can get.

While fleets generally capture only basic information from driver applicants, even this limited data can be used to identify those with a higher risk of leaving early. To reduce that risk, companies could spend extra time with these drivers during orientation training to better manage their job expectations, Croke says.

Omnitracs Analytics has unbundled its predictive models to tailor to the specific needs of fleets of all sizes. Some clients may only want to score driver job applicants on the basis of their likelihood to remain.

Croke also is considering creating a new model that would be offered directly to drivers and training schools. This model would identify the carriers and types of operation – flatbed, tanker, long haul, dedicated, reefer – that would best meet drivers’ job expectations.

Another possibility is to incorporate data about driver physiology such as sleep patterns. Croke is an expert in sleep science and believes that one of the reasons driver turnover is highest in the first 90 days is sleep deprivation while getting used to a new work schedule.

Some of Omnitracs Analytics’ accident models show drivers are at the highest risk at 87 days. Croke anticipates eventually being able to capture the quality of drivers’ sleep from wearable devices that communicate with Omnitracs’ in-cab mobile devices.

Ultimately, the best countermeasure for a driver at risk of quitting may be to make sure he’s able to get a good night’s sleep.

Personality testing

Two years ago, Stay Metrics began a research project with the University of Notre Dame. Seven of the company’s carrier clients and 450 drivers from those carriers provided data for the study.

The drivers completed an in-depth online survey developed by professors Timothy Judge and Mike Crant from the University of Notre Dame’s Mendoza College of Business. The survey was used to assess drivers’ personality traits, and the carriers provided safety scores and turnover data on the drivers throughout the study.

The full study is currently in the peer review process and is set for publication in academic journals within 12 to 18 months. Judge, Stay Metrics’ director of research, already has used the study’s results to create two predictive models for turnover and safety, both linked to key personality traits of drivers.

Each model uses a predictive index based on four personality traits that correlate strongly with turnover and safety. Orderliness is one predictor of driver turnover, while anger is one for driver safety, says Hindes.

“Drivers with an orderly trait are structured – they take notes, make lists and keep their paperwork in order,” Hindes says. “Anger is a personality trait one might expect of unsafe drivers, and when combined with the other traits in the models, a more holistic view emerges.”

Stay Metrics plans to develop a selection tool for carriers to screen job applicants. Field-testing of the two predictive models will begin in December with four carriers whose drivers will take a personality test during orientation meetings. Subsequently, their turnover and safety performance will be monitored for the next six months.

At the conclusion of field-testing, the results will be used to determine the direction of the new selection tool. The earliest this new product would be available is July 2016, Hindes said.

The driver shortage makes it difficult to be too selective when hiring drivers, but by using insights from data to see which drivers are most likely to leave and why, fleets possibly can change the outcome.