Choosing a line at the grocery store can be surprisingly daunting. You can pick the shortest queue and avoid customers with large orders or coupons, yet still feel like your cashier is the slowest. But new research suggests that you may still be on the fast track, provided that line is dedicated to just one cashier.

Previous researchers suspected that a single line leading up to multiple cashiers—the system many ticket sellers and big box stores use—could be maximally efficient. But a forthcoming study in Management Science challenges that assumption. A better system is not a single queue at all—it is many of them. When workers have their own dedicated set of customers, called parallel queues, they work faster.

The findings come from a team of researchers specializing in behavioral economics and behavioral operations at the University of Washington, Syracuse University and Purdue University. Behavioral operations is a fairly new field that has risen, much like behavioral economics, from questioning the basic tenets of their nonbehavioral predecessors—operations management and economics, respectively.

In the past operations management researchers modeled various store checkout scenarios mathematically. The conclusion, generally, was that a single line for multiple cashiers would most efficiently channel people toward the next available register. But these models neglected an important detail: “In operations management, comparing single and parallel queues is one of the most basic things that people have done for a long time,” says behavioral operations researcher and study co-author Masha Shunko of the University of Washington. “But none of them looked at the behavioral aspect before.”

In a series of experiments both online and in person, Shunko and her colleagues created a virtual grocery store checkout line and enlisted a total of 729 participants to act as cashiers. Of those, 248 came to a physical lab to participate while the rest used an online system from all across the U.S. The interface was simple: a box on the left side of the screen showed avatars for the participant and three computer-generated cashiers, each of whom had a dedicated line of customers. The right side of the screen had five sliders representing five variably priced grocery items, which they would ring up by setting the slider to the correct price. The computerized customers lined up in either a single queue or the shortest of the four available queues.

To investigate different aspects of queue design, the authors tested different scenarios, namely single versus parallel queues and a full versus blocked view of the customer lines. The results were the same across nearly every experimental variation: Parallel lines move faster than single ones.

Air Force Institute of Technology operations management researcher Kenneth Schultz, who did not participate in the study, explains: “What this paper is saying is that [compared with earlier research] you’re missing some of the human link.” Real cashiers, unlike mathematically modeled ones, work at different paces depending on how independent their task is, how much feedback they receive and what incentives they are given.

In the new research the authors think the key lies in making workers feel they’re responsible for the line’s progress. Give them their own line and suddenly they have a direct impact on how fast it moves. With a single queue everyone works together to move the line along, which makes each individual less motivated to work quickly. In order to see that impact, though, workers have to have a full view of the line. Towers of merchandise and display cases often block a cashier’s view of the customers. Removing those visual barriers motivates people to work faster because they are getting immediate feedback—they can watch their line shrink. There may also be an element of competitiveness. If a cashier can see how his or her line compares with everyone else’s, a checkout area can become the finish line of a race, with each cashier vying to be the fastest. That competition means faster, more efficient service for customers.

But in a laboratory setting, with its automated cashiers and computerized interface, it may be challenging to suss out the role competition plays in comparison to other motivations. “There’s often a criticism with regard to lab experiments because it’s an artificial environment,” says behavioral operations researcher Elliot Bendoly at the Fisher College of Business at The Ohio State University, also not involved in the research. “But it’s not a far cry from reality in this case.” Behavioral research can be challenging when it’s focused on a subset of people like CEOs, for example, because the subjects in your experiment probably will not act like CEOs. Studying a role like cashier, which lots of people could fill, makes the new research conclusions more realistic and applicable, however.

Although many store managers may not read academic publications, the study authors are spreading the word about their results through the classroom. After learning about queuing theory, students often come back to regale them with stories of inefficient workers and poorly-planned Starbucks lines. That frustration is how the system improves, says Syracuse University behavioral operations researcher Julie Niederhoff, a co-author of the new study. “I love that my students are thinking about how the system could change and do better.”