Labor forecasting requires a stronger analytics skillset

Labor forecasting — or accurately planning schedules — is a key part of running successful shifts and avoiding the negative consequences of overstaffing or understaffing. And the rise of predictive scheduling laws has added a slight bump to the road as managers strive to create schedules up to two weeks in advance. 

Meanwhile, the full potential of analytics in labor forecasting has yet to be realized. In the past year, only 21.1 percent of organizations have used workforce analytics to inform decisions about scheduling shifts, according to the Workforce Business Intelligence Board’s “2020 HR State of the Industry Survey,” developed by Workforce.com’s research team. 

Also read: How to avoid overstaffing through wage tracker software 

Typical companies overstaff, said Matt Stevenson, partner and leader of Mercer’s Workforce Strategy and Analytics practice. But focusing more on forecasting allows the organization to be more fine-tuned in staffing and be more exact in planning their HR spend. 

Labor forecasting by industry 

The complexities of labor forecasting vary by industry, Stevenson said. It’s more straightforward for industries like manufacturing, where the number of workers needed likely depends on how many workers are needed per piece of equipment. Restaurants and retail exist somewhere in the middle of this spectrum and often rely on historical experience or data. 

Health care systems are where it gets extraordinarily difficult, Stevenson said. With a highly specialized job like a neurosurgeon, for example, there’s no way for an emergency room to know for sure if they’ll need one or 10 on staff for the week.

Organizations can attempt to automate this function or analyze historical data, but the reality is that the more accurate and specific someone wants to get with staffing, the more detailed the process has to be.

Also read: Labor analytics add power to workforce management tools

‘Project-ize’ workplace initiatives 

Scott Georgia, vice president of professional services delivery management and transformation at Alight Solutions, suggested that managers create specific projects within initiatives, allowing them to more carefully analyze resource demands for each initiative and estimate the number of hours per week that each initiative will require. 

“This can take some time, but the investment is worth the end result and will allow better visibility into staffing demands, conflicting initiatives and the ability to aid in department priorities,” he said. 

Georgia also suggested that managers gain an understanding of employees’ day-to-day workload, the individual activities that go into that and how many hours per week employees spend on each activity.

Managers can then combine the summarized data points from these first two steps to understand the workload of every employee in the department. They can layer in historical data for further comparison and to help provide data-driven insights.  

“The combined data will provide key insights into what employees are working on, how much they are working on and any staffing capacity or over-allocations,” Georgia said. “Review the data on a monthly basis or more frequently to make data-driven decisions and assignments.”

More sophisticated forecasting 

Labor forecasting has typically been done through the finance department, but that’s starting to change, Stevenson said. Organizations have the opportunity to be more sophisticated in their demand forecasting —  the process in which the historical data or predictive analytics is used to develop an estimate of expected customer demand.

It’s not as black and white as the company needs a certain number of employees to complete a specific amount of work. The type of employee can impact demand forecasting, Stevenson said. Maybe an employee who has worked longer at this job completes tasks quicker or has fewer errors, for example.

In the past, understanding the impacts that different employees have on getting the job done was mostly done by gut, Stevenson said. But now organizations may have this data stored in their administrative systems. To take advantage of this data, it requires access to a system that collects this information and someone with analytical skills on the team. 

The types of organizations doing this typically have been manufacturing companies, since this skill set is generally something that industrial engineers are competent in, Stevenson said. It’s now moving to retail as well, but not as much with health care, whose unique challenges make the shift difficult. 

The makeup of an analytics team

A team or an individual who is able to do this work typically has three areas of expertise or skillsets, Stevenson said. 

The first is the ability to aggregate data from a variety of systems since there has yet to be a single system that can do everything. This is the most basic skill. The second is the ability to understand what the numbers mean and communicate takeaways to the operations team, which will ultimately be the team responsible for the outcomes. The final ability is having specific labor forecasting software or statistical skill to predict demand and labor increasingly better as time goes on.

A lot of software does the job of aggregating the data, but it’s ultimately up to people with specialized skillsets to come up with actionable insights that continually improve forecasting. 

Usually, at most it’s a third of the solution for labor forecasting, he said — not to underestimate the time it can save people by aggregating data. “It will make your job easier, but it won’t replace you,” he said.  

 The toughest part of this for people building this skillset is that it’s difficult to separate data that is interesting from data that is also useful, he said. Teaching that more investigative research skill is not something that many undergraduate schools do, at least not in an analytics program. 

Still, he said she’s seeing more people specialize within these analytics skills. There can be a lucrative career path there, he said. 

Georgia added that to be successful at labor forecasting, a team needs a manager who is curious and who is willing to ask tough forecasting questions. If the current data isn’t available, they’ll have the drive to find the answer.

“This curiosity creates a drive to utilize and visualize the data in creative ways,” he said.