Queues: How Lines Work, Why They Fail, and What Finally Fixed Them

The word is the French word for tail, descended from the Latin cauda, and for most of its life it meant exactly that: a tail, or a braid hanging down the back of a wig. Only in the last two centuries did English speakers start using it for a line of people, which is fitting, because that is roughly when there started to be lines worth naming. Shops, railways and public offices began putting strangers in front of a counter at the same moment, and the queue became the machine that decides who goes next.

That is the useful way to think about it. A queue is an agreement among strangers about order, enforced by nothing but convention and the occasional glare. Every design change since, the rope, the paper ticket, the text message, is an attempt to make that agreement hold with less standing, less doubt and fewer arguments.

From rope line to paper ticket to text message

For most of history the line and the wait were the same thing: your position was your body, and holding your place meant standing in it. The first real break was the take-a-number dispenser, which spread through bakeries, delis and government counters in the mid-twentieth century. It split order from posture. The ticket held your place so your legs did not have to, and you could finally sit, as long as you stayed close enough to hear your number and accepted losing your spot if you stepped out.

The second break happened at theme parks, where the cost of a standing line is easiest to measure: a guest in a two-hour queue is a guest not buying anything. Disney's FASTPASS started handing out return times instead of positions in 1999, and Six Flags rented a handheld that held your place while you roamed. Hold the place, release the person. Once phone cameras learned to read QR codes without an app, the idea stopped needing hardware at all, and a place in line became a record in software, joined from a printed sign and carried in a pocket.

Why lines form at all

Queueing theory began with A. K. Erlang, a Danish engineer studying congestion at the Copenhagen telephone exchange in the early 1900s, and its central finding is uncomfortable: a line can grow even when you have, on paper, enough capacity. No equations needed to see why. Arrivals are lumpy and service is not. Customers do not show up on a metronome; they cluster, after work, after school pickup, after the rain stops. Your service rate, meanwhile, is roughly fixed: one haircut, one consult, one prescription at a time.

When a cluster lands, a queue forms instantly. But it drains slowly, because the only force pulling it down is the thin margin between how fast you serve and how fast people keep arriving. The busier you already are, the thinner that margin, which is why a shop running near full capacity does not degrade gracefully. One bad hour puts it behind for the rest of the day.

The other half of the trap is that service cannot be stored. The idle ten minutes at 9 a.m. cannot be spent at noon. So the levers are few: smooth the arrivals, add capacity you can flex when the cluster hits, or, the one most operators overlook, make the waiting itself cost less. That third lever is where queue design lives.

The four shapes of a queue

Once you start looking, you see the same four designs everywhere, and each one trades something different away.

The single serpentine line

One line feeds every server, the way banks and airport security run it. It is the fairest shape there is: strictly first come, first served, and nobody gets pinned behind the one customer contesting a bill, because the next free server takes the head of the line. The costs are floor space and optics. One long visible snake reads as a terrible wait even when it moves quickly, and everyone in it is still standing.

One line per server

The grocery-store shape. Picking your own lane feels like control, and that feeling is the design's only real advantage. The pick is a gamble: get stuck behind a price check while the next lane empties, and the queue stops feeling like an agreement and starts feeling like a lottery. This shape also invites jockeying, hopping between lanes, which lets service order drift away from arrival order. Drift is where the arguments come from.

The ticketed queue

Take a number, sit down. The ticket separates your place from your feet, which was genuinely radical, and it keeps the order honest without a rope. Its limits are the room and the silence: you must stay within earshot of the caller, the slip tells you nothing about how long you will wait, and the machine keeps no record of the people who looked at the gap between their number and the counter display and quietly left.

The virtual queue

The current step: your place is a record in software rather than a token in your hand. You join a virtual queue by scanning a code, your phone shows a live position and an estimated wait, and a text calls you back when your turn is near, so the wait can happen in the car, the coffee shop or the office. The trade is communication. An invisible line that goes silent is worse than a visible one, so the design stands or falls on honest estimates and a reliable heads-up.

Where queues actually fail

Queues rarely fail because the math was wrong. They fail because people leave them, and people leave for reasons that are psychological before they are arithmetic. David Maister wrote down the classic principles decades ago: occupied time feels shorter than unoccupied time, uncertain waits feel longer than known waits, and unexplained waits feel longer than explained ones. Twenty minutes with a visible position and a moving estimate is a different experience from twenty minutes staring at a door, even though the clock does not care.

Abandonment is where that cost lands. The average Canadian walk-in wait now runs past an hour, plenty of time to change your mind, and when customers give up, the wait itself is the reason they most often give. Worse, the older shapes hide the failure: a sign-in sheet or a ticket roll keeps no record of the person who gave up. Queueing theory has names for these exits, balking for the customer who sees the line and never joins, reneging for the one who joins and quits, and the glossary has the rest of the vocabulary. The operational point is simpler: the shape of your queue decides whether you ever find out.

Frequently Asked Questions