A convenience store manager scans the shelves and has a decision to make. Order more drinks before the weekend rush or play it safe and risk running out. Too many, and cases sit unsold in the back. Too few, and customers walk out upset and empty-handed. It’s a small decision, repeated thousands of times across businesses every day. According to research from Cary Deck, it’s one people routinely get wrong.
Deck’s recent research paper, “Experience Minimizes the Pull-to-Center Effect in Newsvendor Decisions,” published in M&SOM-Manufacturing & Service Operations Management, revisits a classic challenge in operations known as the “newsvendor problem.” This is where businesses must decide how much inventory to order before knowing demand. For decades, researchers found decision-makers consistently miss the mark, failing to account for the asymmetric costs having too much or too little inventory and instead ordering quantities close to average demand. This is a pattern called the “pull-to-center” effect. Even with experience, the mistake seemed to stick.
“What researchers have found in past laboratory studies is sellers tend to pull their inventory toward the mean of the demand distribution, toward the expected value of that distribution, even when that’s not optimal,” Deck said. “It might not be optimal because the cost of running out may be a lot bigger than the cost of excess inventory. Or vice versa.”
Previous studies tried to understand why. Some pointed to psychological biases, like a reluctance to disappoint customers. Others suggested people were “chasing demand” and adjusting orders based on what happened last period.
Deck took a different approach.
“What we did is say, ‘We’re going to let decision makers pick an inventory level and we’re going to play out the demand realization process very quickly,’” he said. “Every few seconds they’re getting feedback about how good their strategy is, and then they can adjust that strategy at any point.”
In this “near-continuous time” environment, subjects in The Interactive Decision Experiment (TIDE) Lab made inventory decisions and quickly saw the profit results, which in turn determined how much money the subject was paid. Instead of a few dozen trials, they experienced hundreds of outcomes in a short span. The difference was striking.
“Our subjects started making more optimal decisions, which is something people had not observed in past newsvendor experiments,” Deck said.
The results suggest the pull-to-center problem isn’t as permanent as once thought. Instead of being a built-in quirk in how people think about the task, it may just come from not having enough chances to learn in a difficult situation.
In slower experiments, a single lucky outcome could reinforce a bad strategy, while a good decision might appear flawed after an unlucky draw. But with rapid, repeated feedback, those distortions fade. Participants begin to see patterns, understand trade-offs and adjust accordingly.
The implications stretch across industries. From major retailers to neighborhood stores inventory decisions shape profits and customer satisfaction alike. More broadly, the study highlights the importance of how people learn on the job.
“What we really show is that the way people are going to behave is going to depend on the experience and the feedback that they’ve had with that setting,” Deck said.
For managers staring at half-empty or overstocked shelves, that insight is practical. Better decisions may not require better instincts, just better opportunities to learn.