Hiring_and_Predictive_Analytics-Selecting_the_Right_Candidates
- 1. Get in touch. Call us at 1.866.963.6941 or write us at info@canworksmart.com. Hiring & Predictive Analytics: Selecting the right candidates to increase driver retention.
- 2. Hiring & Predictive Analytics: Reducing driver turnover 10%. Contemporary Analysis Page 2 Hiring is a tireless search for the best candidates. It requires finding candidates with the values, education, and work experience managers think will lead to success within their company. But, predicting success requires experience. And unfortunately that experience can be unique to each manager. Some of the reasons to hire someone can be verbalized, but most are hard to verbalize — they use intuition. Each manager has a hiring “formula”. Their hiring formula goes beyond tools provided by human resources, because managers ultimately decide who works for them. However, intuition makes hiring formulas hard to share and combine. If hiring formulas can be defined, the hiring process can be improved. Defining hiring formulas allows managers to combine them, test them, and track their results over time. Companies have struggled to do this because they can’t capture the subtle (and less–subtle) reasons for hiring or not hiring a candidate. Thankfully, predictive analytics makes this possible. Predictive analytics can create a hiring formula using data from every applicant and employee — and their job performance. Predictive analytics uses data to capture the patterns in your company’s collective experience, across employees and time. Don’t worry, you likely have the necessary data. All companies, except the newest, have the data necessary for predictive analytics. We work in the digital age. Companies collect data on everything. How would your interviews change if you could focus on only qualified candidates?
- 3. Hiring & Predictive Analytics: Reducing driver turnover 10%. Contemporary Analysis Page 3 Every event and action is translated into data. The key is knowing how to find the data and unlock the right patterns. The value of data is in the patterns. The patterns in your data allow you to predict the future and describe the past. In this case study we explain how we used the patterns in a trucking company’s data to tell them who to hire and why. CAN used predictive analytics to reduce driver turnover 10% and save $1.7M per year. The problem CAN was approached by a trucking company that hires 5,000 drivers a year. They had an annual driver turnover of nearly 100%. Unfortunately, high driver turnover is common in the trucking industry. High turnover wasn’t the problem: the problem was drivers were leaving too early: before the company could earn a profit on hiring and training each driver— an investment of $3,500 per driver or $17.5M in total per year. The following chart shows the distribution of the total number of weeks every driver in the study was employed. Number of weeks it takes each driver to break-even
- 4. Hiring & Predictive Analytics: Reducing driver turnover 10%. Contemporary Analysis Page 4 The company broke-even on most drivers between the 10th and 30th week of employment, depending on the number of miles they drove. However, this was also when the probability of drivers leaving increased most dramatically—as illustrated above. CAN was asked to identify ways that the company could reduce the probability of drivers leaving between the 10th and 30th week. The result was an increased return on the investment in hiring and training drivers. The solution This problem was different than CAN’s other employee retention projects in retail, insurance and banking. Typically we identify employees that are most likely to leave, allowing managers to give “at risk” employees more attention, new assignments, encourage them, or make them feel special. During the discovery process for this project we decided that this approach was unlikely succeed in trucking. Instead of trying to improve employee retention post-hire, we decided it would be more effective to hire drivers that had higher probabilities of staying. We decided to improve the hiring process, instead of the management process. Over 3 years $35,639,000 was invested to train ~10,354 drivers who quit within one month of completing training. How can this slope be flattened? Probability of being employed by week
- 5. Hiring & Predictive Analytics: Reducing driver turnover 10%. Contemporary Analysis Page 5 CAN started with 3 years of data, and after cleaning the data, had reliable data on 12,819 drivers. We divided the drivers with 9,701 in the study group and 3,118 in the control group. Our analysis looked for patterns in the following variables: The results CAN’s analysis found 6 variables to be statistically significant predictors of driver retention—3 are controllable. We made recommendations based on what the company could control and presented decision makers with estimated impacts to driver retention and savings. The following is a discussion of our findings for all 6 variables, including our recommendations. 1. Hire type This variable is uncontrollable. The company hired several types of drivers, however the majority all fell into one hire type. They have little control over this, even if other hire types have higher retention. 2. State This variable is uncontrollable. State is most likely an indicator of local competition from other trucking companies and competing industries such as construction. While our client can avoid hiring some drivers in highly competitive areas; in the long-run they can’t avoid hiring drivers where they are needed. 3. Training school This variable is controllable. The company hires drivers from several thousand training schools—5 training schools have higher than average driver turnover. Our recommendation is to avoid hiring from these 5 training school: saving the company $86k per year. The following graph shows the probability of driver retention by week before and after the 5 training schools were removed. 1. Gender 2. Training school 3. Hire type 4. Number of unloads 5. Family type 6. State 7. Age at hire 8. Average net pay per mile 9. Eligible for rehire or not 10. Average net pay per week 11. Days tractor broke–down 12. Home time requests rejected
- 6. Hiring & Predictive Analytics: Reducing driver turnover 10%. Contemporary Analysis Page 6 4. Age at hire This variable is controllable. Age is a proxy for experience, CAN recommends focusing on experience when possible instead of age. However, age at hire is a statistical significant predictor of turnover. The following graph shows the inverse relationship between average turnover and age at hire. Average Turnover by Age at Hire Weeks Employed %Retention 60% 40% 20% 80% 100% 0 5 10 15 20 25 30 35 40 Before After Removing the bottom 5 training schools Impact of $86k per year
- 7. Hiring & Predictive Analytics: Reducing driver turnover 10%. Contemporary Analysis Page 7 We wanted to understand why age at hire had an inverse relationship with average turnover. We interviewed several drivers and managers. We discovered that many people looking for work are told that they should go into trucking and apply without understanding the job. We also discovered that more experienced workers with higher retention were more likely to come from jobs such as farming that required long hours of driving. While many less experienced workers with lower retention came from construction, manufacturing and the military. The theory and data supported each other — the key to good data driven research. The company needed to continue to hire qualified drivers of all ages. CAN did not recommend the company only hiring drivers of a certain age, but that they change the distribution of drivers. The following graph shows the distribution of drivers by age at hire that CAN recommended. Current and Recommended Workforce by Experience Before After 36-40 41-45 46-50 51-55 56+ Age at Hire 31-3526-3020-25 10% 5% 0% 15% 20% 25% %ofDriverPopulation
- 8. Hiring & Predictive Analytics: Reducing driver turnover 10%. Contemporary Analysis Page 8 Shifting their workforce to favor experience helped them save $358k per year. Weeks Employed 20 25 30 35 40151050 Before After20% 40% 60% 80% 100% %Retention Adjusting Workforce for Experience Impact of $358k per year 5. Average net pay per week This variables is controllable. Drivers are compensated by how many miles they drive. We found that if the company could provide drivers with an additional $70 worth of work a week they could increase retention for a savings of $1.3M per year. Weeks Employed 20 25 30 35 40151050 %Retention 60% 80% 100% 40% 20% 0% Before After Increasing Work and Pay $70 per week Impact of $1.3M per year
- 9. Hiring & Predictive Analytics: Reducing driver turnover 10%. Contemporary Analysis Page 9 6. Number of days tractor is broke-down This variables is uncontrollable — beyond regular maintenance and safe driving. However, breakdowns means drivers can’t earn a living. Drivers are assigned a tractor and are paid by the mile. If they don’t drive, they don’t get paid. It was not surprising to find breakdowns as a significant driver of driver retention. Conclusion In the end CAN helped the company reduce driver turnover 10% and save $1.7M per year. Weeks Employed 20 25 30 35 40151050 Before After20% 40% 60% 80% 100% %Retention All 3 Recommendations Impact of $1.7M per year This case study show that predictive analytics excels at helping companies understand and manage the complexity of hiring employees — from defining selection criteria to filtering applicants. Predictive analytics allows companies to capture the experience and intuition of your company and managers to demystify human resources and create a shareable hiring formula. We look forward to helping you increase your employee engagement and reduce turnover by hiring the right employees.
- 10. Get to know us. Learn more about CAN. Contemporary Analysis Page 10 Our solutions are used by fast-growing technology companies, Fortune 500s, as well as small- and medium-sized organizations. Our clients are in a variety of industries including construction, insurance, education, healthcare, government, not-for-profit, software and engineering. Our vision is to make predictive analytics simple and affordable because all companies, not just the largest, should be able to benefit from predictive analytics and data science. Our principles: 1. We care about business. Each business deserves a custom solution. Problems are our passion. 2. We solve core business problems. We make a big impact quickly. Value is our focus. 3. We don’t have all the answers. We help our clients make better decisions. Less wrong is the goal. 4. We are technology agnostic. We focus on the solution. Technology is just a tool. 5. Our job is to solve problems, not introduce complexity. Our solutions are simple because our clients are busy. Since 2008, Contemporary Analysis has used predictive analytics and data science to help companies of all sizes work smart. Our five products use data to help our clients improve their sales, marketing, customer service, management, and strategic plans.