Science of queuing for minimum wait in line

Just in time for the traditional kick-off to the holiday shopping season: here’s an interesting essay via The Conversation, on why a single long line, served by multiple cashiers, can be more efficient and move more quickly than several smaller lines each with a single cashier. The essay focuses on grocery stores relying on the “single-server” queuing model — but I’ve observed that many grocery stores combine single-server and multiple-server models.

Individual cashiers continue to each ring customers through individual lines, and choosing wrongly can result in significant wait. But in the same store, a single line of customers will form for several self check-out devices. And sure enough, even if one self-module breaks down or is subject to slow-down, that wait time is dispersed across the entire system. Customers can move to the next available unit instead of having to wait for that one.
​
From experience, I’ve found that multiple-server breaks down when a single cashier is assigned permanently to the registers and is then tasked with summoning additional personnel as-needed. Inconsistent policies defining a “long line” among various managers, piled-up merchandise blocking the view of customers queuing in line, can undermine the so-called “Lead Cashier” in the performance of this task. Too easily, this person ends up becoming a scapegoat for customer dissatisfaction.