Long hold ties can risk a customer's patience—and be costly for business. Companies that use call centers may frustrate or lose waiting customers, in part because they have an incomplete understanding of callers' patience levels and patterns.
Research by Chicago Booth Professor Bariş Ata and Chicago Booth Associate Professor Che-Lin Su, conducted with Koç University's Zeynep Akşin and the University of North Carolina's Seyed Morteza Emadi, aims to correct that with a first-of-its kind model that analyzes caller patience. Their findings create faster, more efficient call centers.
“Knowing when a person will decide to hang up or hang on is vital to streamlining call-center operations, minimizing caller frustration, and maximizing every customer-service encounter,” says Su.
Past research into call-center operations treated caller patience as a constant, unchanged by how well call centers routed or prioritized calls. Such assumptions made it simpler to analyze the data and apply it in the design of call centers. But the assumptions ultimately failed to provide a reasonable picture of people’s willingness to wait on hold, which could be at the root of poorly run call centers, according to Su.
Su and Ata wondered if it was more likely that callers’ tolerance would change depending on how likely they were to get the service they wanted, and whether caller patience could be influenced by how a company chose to handle calls. “When a call center alters its protocols to improve speed and service, adds agents, or changes call routing and priority, we theorized those changes should influence caller patience,” says Su.
The researchers present a dynamic model of the caller-patience-and-decision process, based on data drawn from 1.3 million calls made between April and September 2008 to an Israeli bank’s customer-service center.
Each caller put on hold must decide whether to abandon a call or continue to wait, and the researchers model wait or quit decisions and estimate callers’ patience using data on waiting and abandonment times.
To test their theory, and compare it to older models, the researchers applied four scenarios that simulated changes in call-service discipline. Those included a first-come, first-served policy, in which calls are answered in the order they are received; a static policy, in which the highest priority calls are answered first; and two “threshold” policy scenarios, which both adjust call-handling priority based on one of two preset time thresholds for how long callers have to wait.
The results reveal that the older models’ assumptions were at times off substantially. They overestimated how long a caller was willing to hang on, for example—6.6 minutes versus just under 90 seconds—and underestimated a caller’s tendency to hang up by more than 5%, the researchers say.
Predicting caller patience more precisely can help businesses design better call-center systems, fine-tune those currently in use, and negotiate smarter contracts for outsourcing such services. Because the model better predicts how long a caller will stay on the line, it enables more precise estimates of how many calls can be served per hour, day, and month, says Ata.
More precise caller-patience estimates are important when companies change their business practices or launch marketing promotions that can produce call surges. Ata notes, “It’s no use spending millions of dollars advertising a new product, service, or event if your call center can’t cope with the customer response.”