Field service is full of in-day surprises that can lower efficiency and cause SLA violations and poor customer experiences. To minimize the amount and impact of disruptions, it’s important to have an accurate schedule and ensure each job is assigned to the appropriate person. Predictive field service, powered by machine learning and data science, is key in making this happen.
An important part of predictive field service is accurately estimating job duration. And it’s not as simple as using the average time it takes to complete that job type. There’s more to consider, such as required skills and technician efficiency, to start. Predictive job duration estimates the most precise time it will take to complete a job based on all relevant task and engineer properties. And it continuously learns and improves from historical data. This means, for example, that as a technician gets more efficient fixing a specific issue, the expected job duration will shorten over time.
When used correctly, predictive job duration can be highly valuable to field service organizations—it can lead to greater schedule accuracy and greater workforce efficiency. But as when implementing any new technology, it’s important to be cautious. Before businesses can capitalize on the value, there must be a new set of standards to replace the old ways of assigning job duration.
Here are some of the best practices that field service organizations can follow to maximize the value of predictive job duration:
1. Involve Your Field Resources Throughout the Process
First, it’s important to fully disclose to your employees (technicians and engineers especially) how your organization will be using predictive job duration. Then, emphasize how it’s meant to help both them and the entire organization. Assure your employees that you aren’t setting them up to fail or make their jobs more difficult. Let them know you trust them not to take advantage of the situation either—because it can happen.
For example, if Field Tech Frank realizes that his assigned job duration is elastic, he might work slower so the field service management system gives him 40 minutes for future jobs instead of the 30 he actually needs. This gives him an extra 10 minutes for a coffee break.
Meanwhile, Field Tech Joe might be getting better and faster at completing a job type, so the system gives him less time in the future to complete a job. Now he’s working harder than Frank, yet it’s harder for him to meet SLAs. He now gets flagged when taking more than 25 minutes to complete a job, while Frank takes 40 minutes and there’s no problem.
Now there’s no incentive for Joe to continue to work quickly. Not to mention, your organization is unintentionally rewarding bad behaviors and penalizing those who are improving at their jobs. To avoid this, always be monitoring for any abnormal behaviors in job duration. If something falls outside of the standard variation, talk to your team and be there to listen to any of their concerns.
Likewise, while the overall objective of predictive job duration may be to be improve business operations, it’s not just beneficial for the executive team. It’s also a way for technicians to learn about trends in their performance, uncover personal strengths, and recognize areas for improvement. Be sure to make this benefit clear to your team.
2. Compare Field Technicians Against Themselves
Managers and supervisors can also use predictive job duration to determine how to best use resources and make better assignment decisions. By monitoring individual tech performance and comparing their own performance across different locations, times, customers, job types, and so on, you can discover the optimal situation where a specific tech is most efficient. Likewise, you can uncover where a tech is struggling, and work towards improvement.
For example, let’s say your field service management system learns that Joe takes 30 minutes to replace a meter at 10:00 am Monday and 50 minutes to do the same job on Thursday at 4:00 pm. You can determine that Joe performs better in the mornings and earlier in the week. From there, you can adjust accordingly to ensure that Joe is utilized effectively in any scenario. This could mean:
- Joe’s supervisor monitors him more closely on Thursday afternoons
- Changing Joe’s shifts so that he’s assigned jobs for the morning when he tends to perform better
- Assigning Joe low priority jobs during situations when he’s less efficient
- Asking whether other factors are playing a role—travel time, weather, or the likelihood of a customer being late for the appointment, for example
3. Analyze Company Performance
It’s easy to focus on individual technician performance when measuring job duration—but it’s not the sole purpose. It’s also important to go beyond looking at job duration at an individual level and maintain a holistic view into operations. With full visibility into performances across regions, job types, customers, and field resources, you can uncover which areas are strong and which areas need improvement. This then allows you to set standards, define goals and objectives, and work towards improvements to the organization as a whole.
It’s one thing to learn that Joe takes 20 minutes less than Frank to replace a meter and should be dispatched to that job. But there’s much more to uncover, such as why all jobs with a certain customer are taking longer than expected to complete. Or why one region has the longest duration for a particular task type.
Bottom line: if you’re only using predictive job duration to improve technician performance and not to gain a holistic understanding of organizational processes and field resource dynamics, you’re missing out on valuable information.
4. Keep Strong Data Records
Finally, any machine learning-based predictive field service solution requires high-quality historical data (and lots of it) to accurately forecast the future. This means collecting structured data like technician skills and job history, weather and traffic patterns, first-time fix rate, customer demographics, parts usage, and of course job duration. It also means collecting any notes from past visits such as about issues that occurred on the job or customer feedback. Moreover, it’s crucial to keep adding to this data to continually improve predictions.
If you already have a strong data set, you can train, test, and validate predictive capabilities like job duration, and get a head start in improving business processes.
Ultimately, whenever changing a process or implementing new technology into a business, there will be challenges and you will need to redefine standards. It’s important to involve the entire organization in the process, especially if it directly affects employees. And in the case of predictive job duration (or any predictive capability), the best approach is to measure results from both a holistic and individual technician view—while continuously improving the process with data. As long as it’s used correctly, predictive job duration will deliver the expected business results and improve your field service operations.
For more field service management best practices, visit the ClickSoftware blog.