Thursday, December 04, 2008

A note about queues and service windows

We've all experienced that annoying sensation that we're in the "slow queue". The line next to us seems to be moving faster, but you know that if you shift queues that'll be the when the new line slows down. You've also probably seen that some places set up their queues differently. Typically, you'll see one of the following configurations:
  • everyone queues up in one line, and goes to the next open service window/register/person (often seen in banks these days, and at airport check-in counters); or
  • people join a queue behind one service window or another. Typically, people will join the shortest queue. (This is what you typically see in fast-food outlets and supermarkets.)
What you may not know, is that getting people to form a single line and go to the next available service window is much more efficient for everyone. Because no-one is necessarily held up by the person with all the excess baggage, or the wheelbarrow full of pennies that need counting, everyone tends to get served faster, on average, than the other model.

The down-side, is that the single queue *looks* longer, and can increase the potential for people baulking. But, the queue also moves much faster, so people are also less likely to abandon the queue altogether.

When designing your service, and looking at your layout options at check-out, it's important to keep these things in mind.


cafedave said...

I guess the baggage check-in is the best example of this. There are two extra complexities that come to mind, though: the extra time spent travelling along the winding queue, and the way to be called to the front of the line if your flight is about to leave. The latter system penalizes the good customers: perhaps the one queue system needs some more tweaking?

Steve 'Doc' Baty said...

Dave, not so. The queue modelling looks at the time in the queue & the time to get served. Total time in the system is optimised with a single queue - including all that time shuffling forward :)

As to the second point - where they call people forward who would otherwise be late for their flight: how is that different from calling those people out of multiple queues? The others are still being penalised.


Tom Voirol said...

Great subject, Steve, and one I've given some thought to in the past.

See, I'm not convinced the single queue makes proceedings any faster. My argument is based on the premise that the time each person spends at the service window is not constant but can be shortened or lengthened by the person being served (think "oh, yes, while I'm here one more question...").

If you're at the window with three people queuing directly behind you (breathing down your neck), you know you're wasting the time of an actual, concrete person (that scary bloke with the mo) by faffing around.

If there is no queque behind you, your taking longer than you need to feels like it only indirectly holds up any person from the faceless crowd which, at any rate, is seated comfortably (in the case of my local bank).

Not sure how pronounced that effect is, but I'm pretty sure you can't entirely discount it.

Steve 'Doc' Baty said...

Tom, that's an interesting idea, but the aggregate effect is an improvement in the overall time for people queuing.

Now I know that mathematical modelling is not the same thing as real human behaviour, but I've seen studies from observational data taken in banks when single queues were first being tested (via CCTV so the customers didn't know they were being filmed) and the results were conclusive - and match the modelling predictions.


flipsockgrrl said...

The mathematical modelling seems to apply in the real world -- I'd guess the "waiting time" displays for queues at places like Universal Studios (Los Angeles) are based on a standard formula involving the number of people waiting and duration and capacity of the ride/theatre show.

(I've only seen this system in action in the USA. Presumably the Australian movie theme parks have similar queue management systems...?)