When long queues can be better than short queues

Joost Vles writes about a subject that I have long had an interest in and that is queuing theory. I have noticed that airports and banks tend to favor the single line system but grocery stores go for the separate queues. He points out what should be obvious but that some people do not seem to realize, and that is that a single long queue where the people break at the head of it to go to the next available server is better than each server having their own queue.

Seeing a line snake back and forth across the width of a store three times can be deceiving as to how long you may actually have to wait. In what may appear to be a very long line, the service rate can be so good that the line moves very quickly.

Imagine a situation where you have many shorter lines, each being served by its own cashier. Call it the grocery store model, or the single-server model, more officially. You can get out of there quickly only if you correctly guess which line will move the quickest. And if you’re anything like me, you’re bound to bet on the wrong line.

But a single, longer line, being served by multiple employees – think banking, the motor vehicle department or airport security – is actually faster for everyone, even though it looks much longer than what you’re used to seeing in other systems.

The main reason is that if there’s a price check, a return or some other very slow customer, that delay affects only that cashier directly dealing with the situation. The rest of the line continues to move along. The delay at one cashier gets distributed across the entire system in the multiple-server model, instead of completely stalling out just that one line, as in the single-server model we see in the grocery stores.

Vles says that designing the most appropriate queue is based on Little’s Law that states that “over time, the number of customers in a system is equal to their rate of arrival multiplied by the average time they spend in that system” and that some systems (like mechanized car washes) have fixed times of service while others like the post office have varying times.

The place you are most likely to encounter the single-server model is the grocery store, perhaps because the layout of the store is not conducive to having a single long line, even if that is more efficient. But another factor may be that people get annoyed by seeing long lines because they do not appreciate that it is actually more efficient. As a result, when one goes to a grocery checkout, one has to guess which line is likely to be the quickest. It really does not matter that much since the difference in times between different lines is not that great but there is something annoying about joining a line that seems the shortest and then getting held up because of a customer ahead of you while the neighboring longer line zips along quickly. It feels like losing a bet.

Over time, I choose grocery store lines using a rule of thumb that is blatantly ageist and sexist. Other factors being roughly equal, I tend to join lines that have only men, the younger the better. This is because men tend to load their stuff on the conveyor belt and pay whatever they are told, no questions asked. They never seem to notice if the price charged for an item is different from what it was displayed on the shelf or was advertised in a store flyer and ask for a price check. When it comes to payment in cash, they never try to pay the exact amount by digging through their pockets to find the exact change but simply give large bills and not look too carefully if they get the right amount of change back. They never write checks. They never have coupons. I doubt that they would notice even if an item they purchased was not put into their shopping bag. Women, especially older women, tend to be much more observant and careful shoppers. As a result, lines of men move faster than lines of women.

I have of course not done any research on this subject and am generalizing based on my own experience, never a wise thing. For example, all the male characteristics I have listed above are strongly based on my own behavior. Sweeping generalizations are course never true in every case and my personal impressions may be too restrictive or only representative of my older generation.

But since grocery shopping is something that everyone has to do, I am curious as to what readers think, whether these generalizations have any resemblance at all to their own experience.


  1. lanir says

    I’m one of those people that when I play games, I always focus on what I’m doing and if there is one, what the board looks like. I don’t focus on how my opponents are playing even in games that pit us in direct competition. So I just pick whichever line seems to be short or has carts that aren’t as full. And then I frankly just try not to be the one holding up the line.

    I actually prefer shopping at night when there are less people. You have to walk around the employees who are restocking but it’s otherwise much more convenient. I feel like I’m in the way a lot less than I do in a crowded store where I have to watch out for blocking an aisle or parking my cart in front of products someone else wants to pick up while I decide on a brand/flavor/whatever.

  2. Bruce H says

    I’m with lanir. If you’re in a hurry, pick times when the grocery store is not busy. Most of all, do NOT go to the grocery store between noon and seven p.m. on a Sunday.

    Of course, I, myself, plan to do my Thanksgiving shopping tomorrow. Most likely about one or two in the afternoon. Because that’s the most convenient time for me, for seemingly everyone else in my town, and because I’m a glutton for punishment.

  3. janicot says

    I always figured it was additional shelf viewing space for the stores. Banks and libraries don’t have a good reason to force customers to look at impulse purchase inventory.
    If it makes you feel better to spend energy trying to save a little time trying to choose a line, knock yourself out. I assume it gives you some satisfaction (or distraction) during a chore.

  4. DonDueed says

    The single-line model wouldn’t work well for grocery stores. The issue is the fact that grocery shoppers have a cart full of items that must be placed on the conveyor in advance of the cashier’s attention (i.e. while the previous customer is still completing the transaction). Yet most delays occur during that last step (price check, making change, etc.)

    So even with a single line, you could still end up stuck behind a slow customer, but you’d lose the advantage of unloading your cart before the cashier was finished with that previous customer.

    Besides that, the stores wouldn’t be at all pleased to yield a lot of space at the storefront for a winding line of carts, since that space would have to be there at all times (even when less busy).

    If anything is going to change the multi-line model at groceries, it’s the promise of checkout-in-the-aisle as now being tried out by Amazon (IIRC). No checkout lines at all!

  5. robert79 says

    One reason for the grocery store model is psychological. People prefer situations where they have (the illusion of) control, and when they have this they to place blame more on themselves than others. In the case of the grocery store this results in people blaming themselves for the delay, Vles does this when he says “And if you’re anything like me, you’re bound to bet on the wrong line.” This results in less cashiers getting yelled at.

    Another reason for the grocery store model is when the time cost of your cashiers is cheap. When the store is quiet a multiple line queue has greater odds of having cashiers doing nothing. If the cost of doing this is low (either because labor costs are low, or because your using automated cashiers, as is the case where I live) this wasted time-cost can be balanced against the benefits of helping your customers faster.

  6. Michael Sternberg says

    I think DonDeed is right in pointing out the time for cart unloading to nudge single-line checkouts towards being infeasible for universal application.
    I have, however, often seen single-line being used for a cluster of rapid checkout stations (“10 items or less fewer”). For those stations, the time for cart unloading would be negligible, bolstering the argument.

  7. says

    My local supermarket has three kinds of checkout, the traditional parallel type for big trolley loads, two self serve ones and three serial ones for basket loads. As mentioned above I too tend to avoid doing the big shop at busy times so I don’t have to queue forever. The self serve ones make their plaintive “welcome” noises as I head by for the serial checkouts. There the shorter evening queue is dealt with efficiently by two or three attendants.

    The other game is to pretend that multiple packets of the same type (favourite items on sale) count as one item so you can use the serial checkout with their ’12 items or less” signs.

    Somehow you have to make menial tasks a little more interesting.

  8. says

    The single queue only works if transaction points are fully manned at all times, and they never are.

    For example, let’s say I go to the bank to make a withdrawal. I go in, get in line, notice that there are only three tellers working when the counter should be at full capacity (7 tellers), immediately slowing progress.

    Further, I often see one or more tellers closing their windows in the midst of the rush, thus reducing the number of available tellers to one.

    During the busiest hours of the day. (Lunch hour and after-work errand rush.)

    With the queue going — no lie — nearly out the door.

    That’s not just inefficient, it’s bad customer service.

  9. Jenora Feuer says

    I remember seeing discussions of queuing theory back when I was getting ready to go to University in the mid-1980s, in particular this issue; at about that time a number of banks and the like were actively transitioning to the single-line model because of some of this work. (This was also the time when ATMs were really starting to come into play, so redirecting people with simple issues out of the line was something that was just starting.)

    I can see the point that the single-line model wouldn’t necessarily work for grocery stores because of the unloading time and the fact that you want overlap between customers at a given checkout to maximize movement. But that same argument applies to airport security checks, and most of those that I’ve seen have gone for a hybrid model with one long line leading into a bunch of short lines which are kept at only a few people each. (With the Nexus card being the equivalent of the self-checkout for diversion, I guess.)

    I suspect the main things stopping grocery stores from going to that model (aside from ‘but we’ve always done it this way’) are things that the airport security check uses to deal with this: dedicated space for the main line, controlled access to the shorter lines, and dedicated personnel just for queue management. Those would cost money and take away space from showing off things to sell, all for events that aren’t happening all the time. Barring a really convincing argument, I don’t think you’re going to see your standard grocery store switching over.

  10. NORM REMP says

    The Toys R Us in my town converted to the single line concept (do not know about the others). It had a little bit of the “unloading” problem, but not nearly as bad grocery stores and I thought it worked great. Over all, it took about the same amount of space, but everyone’s time in line was shortened. In busy times, they had a person directing the next customer to the open register.

    What irks me about grocery store lines is when the line gets long and a new line or two opens, all the people at the back of the queue rush up to the new open register. The people who have been waiting the longest still have to wait and leave later than the people who jumped to the new lines.

    I think there are ways to handle the unloading problem, but it would cost additional capital, so no one will convert unless a major retailer does it and is successful.

    The long line concept would actually allow them to display a wider variety of impulse items, where as the multiple line methods only allows them to display a small number items multiple times.

Leave a Reply

Your email address will not be published. Required fields are marked *