Analysing behaviour at the checkouts is the basis of effective queue management. Customers loathe queues, and avoid retailers where queuing is common. If a supermarket declares a maximum queue length metric of 1+1 (in other words, one person or family group checking out, with one further customer behind), it can immediately open a new till when this figure seems likely to be exceeded. Conversely, when queues are short, checkout staff can be redeployed to other areas of the store.
Most Irisys queue management customers use a two stage system. The first stage uses a 30 minute forecast of checkout arrival rates based on the previous ten weeks of historical information. Monday 10am rates tend to be similar, for example, as do Wednesdays at 4pm. The second stage is to apply live people counting data to ensure the system works accurately in real time.
Customer satisfaction
While the people counting technology may be relatively simple, the queue management software is extremely complex. It has to recognise that a group of six people may be separate customers, or a large family completing a single transaction. It has to understand that a certain percentage of customers will temporarily abandon their trolley at the checkout to race off for a forgotten deodorant, and it has to account for variations in the weather, paydays, holidays... the list is almost endless.
“When the software gets it right, the supermarket greatly increases customer satisfaction rates, and simultaneously cuts the cost of customer service,” notes Galloway. “And, because the software is adaptive, it continues to improve its accuracy as time goes on.
“That’s the essence of great R&D: solutions that not only lead the industry, but get better all the time.”