Long lines at check out can spoil a shopping trip just as a bad dessert can spoil an otherwise fine dinner. Either can, if you will, leave a bad taste in your mouth. So what can a retailer do besides throw (expensive) bodies at the problem?
As the Wall Street Journal tells it, there are quite a few options. A recent article discussed process changes and new technologies different firms are using to try and reduce customer waits (Retailers Wage War Against Long Lines, May 2). The most interesting to my mind was what supermarket chain Kroger is trying.
"Supermarket giant Kroger Co. is winning the war against lengthy checkout lines with a powerful weapon: infrared cameras long used by the military and law-enforcement to track people.
These cameras, which detect body heat, sit at the entrances and above cash registers at most of Kroger’s roughly 2,400 stores. Paired with in-house software that determines the number of lanes that need to be open, the technology has reduced the customer’s average wait time to 26 seconds. That compares with an average of four minutes before Kroger began installing the cameras in 2010.
“The technology enabled us to execute at the front of the store without that additional (labor) expense,” said Marnette Perry, senior vice president of retail operations for Kroger.”It’s remarkable that we’ve been able to improve execution as much as we have without a big price tag.” …
The system includes software developed by Kroger’s IT department that predicts for each store how long those customers spend shopping based on the day and time. The system determines the number of lanes that need to be open in 30-minute increments, and displays the information on monitors above the lanes so supervisors can deploy cashiers accordingly."
So Kroger isn’t so much using technology to substitute for labor or to make labor more productive as it is making sure that its staffing is tailored to actual demand. This is a pretty neat idea. Stores have long been able to use traffic counters to build a forecasting model of customer arrivals. I have seen demos of fancier systems that promise to be more accurate than simple electric traffic counters. For example, they may be able to differentiate between two adults walking in and a child trailing a parent.
But in, say, a grocery store, that is only half the battle. At a call center forecasting how many calls will arrive in some time bucket is sufficient since (in a perfect world) customers move immediately into service upon arrival. In retail, you also need to estimate how long people will shop since customers aren’t ready to move right into check out after walking through the door. That’s where the cameras above the registers matter. Otherwise, one has to estimate the time of arrival at the register from check out receipts but that builds in wait time at the registers. That is, one could estimate that a spike in arrivals translates into a spike in check outs 25 minutes later. But if you are just looking at register receipts, you don’t know whether customers are shopping for 24 minutes and waiting in line for only one minute or shopping for 20 minutes and waiting for five. With the cameras, one should be able to build a decent model of how long shopping trips are and how they might vary by day of week and time of day.
The question then is how to deploy staff to take advantage of this information. Compare this to a call center. In a conventional call center, they have a forecast of how many call will come in on, say, Tuesday morning. On any given Tuesday, that forecast might be wrong. The call center manager might realize that but there is not much the manager can do. It may not be possible to call in extra staff if the actual number of calls significantly exceeds the forecast. The front-end manager at your local Kroger might actually be in a position to do something. Since there is a time offset between when customers enter the store and when they check out, he or she has a bit of time to staff up the registers — assuming that there is flexible labor in the store to pull from stocking shelves.