Quirky Queueing

by vpundir | September 1st, 2007

I followed my wandering feet to High Street Kensington today. And there it was: A shiny new Whole Foods store!

In case you are unfamiliar with Whole Foods, it is a US-based retail chain of natural and organic foods. In fact, it is the world’s largest such chain and recently acquired its closest competitor Wild Oats.

While a Whole Foods is a relatively familiar sight in US metros, I had never seen a store in London before today. The large display windows were decorated with freshly baked bread, complete with a folksy wooden cart to…well, cart them around.

It was around lunchtime, so I popped in to help myself at the hot bar. I was only slightly surprised that hot bar was even more strongly dominated by Indian recipes than the Whole Foods hot bars in America, with “Indian food” being as popular as it is in this country.

So I got some daal makhani from the hot bar, and some cous cous, asparagus, roasted garlic, beetroot, and chick peas & aubergines from the salad bar and proceeded to the check-out.

That’s when I noticed something amiss. The queue seemed way too long. To my pleasant surprise though, it was moving relatively fast. And as I got to the rear-end of the queue, I could see why.

Instead of the parallel tills that one sees in US stores, or even in London groceries, Whole Foods in High Street Kensington has opted for the sequential approach that one would normally run into at London banks, post offices, and underground ticket/information offices.

Here’s how it works: customers form a single queue, and a ticker/public-address-system tells the next customer which till/counter they should go to and when. The rationale, I guess, is fairness. The powers that be decided, at some point in time, that customers can not be trusted to maintain equilibrium if separate queues are used for each window. Though the rationale can be challenged using game theory (not to mention the fact that self-selection makes it relatively envy-free), this method works reasonably well where it is used because there are typically between 2 and 4 windows/ tills/ counters. Additionally, in some of these 2-4 counter places, this queueing can be justified due the lack of floor amplitude.

But here at Whole Foods, this makes absolutely no sense. In fact, if they tried to find a worse way of queueing, I doubt that they’d be able to find one. Here’s the thing: they have 29 tills. While these are wrapped around the hall, giving an impression of three different banks, it is actually just one large bank of tills catering to a single queue.

Here’s the problem with this method: A customer takes (say) 2 seconds to walk to counter 17 (the counter closest to the head of the queue), and (say) 35 seconds to walk to counter 29 (the counter farthest from the head of the queue. Even if the two wraparound legs are not open, the customer would take (say) 20 seconds to walk to counter 1 (the counter farthest from the head of the queue if counters 18-29 are closed). Essentially, there is a difference of at least 18 seconds between counter accessabilities of the first and the last counter.

This can create several problems for employee management. For one, if the employees are being assessed on the number of clients served per day, then the servers at the last counters will (unfairly) appear to be less efficient. If they are not being assessed on this sort of false efficiency, there is the danger that since the end-counters ensure more “downtime”, these positions are much envied among servers.

Most importantly, the whole thing is highly inefficient. On a day when all counters are working (a schenario that they no doubt hope for), Whole Foods is wasting 33 seconds per customer at counter 29, 32 seconds per customer at counter 28 and so on. This quickly adds up to a few full-time equivalents.

So one is left wondering – “what were they thinking?”

Comments are closed.

About This Site

Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Aenean commodo ligula eget dolor. Aenean massa.