I went to the post office at lunch time. Lots of other people did too, so it was quite busy.
For the last few years, this post office has had a fancy queuing system (though it now no longer seems to require a dedicated member of staff to explain the self service options, which is progress of a kind). I took my ticket and prepared to wait.
But then I realised that it might be quite a wait. My little ticket told me that there were 54 people ahead of me in the queue, which sounded like a lot. There were three counter positions open, so if each transaction took about two minutes, I could expect to be there for half an hour. A moment’s further thought suggested that there was a big unknown in that estimate. An obvious reaction to facing a queue 54 people long is to go away and come back some other time when the queue is shorter. But it’s only possible to find out the length of the queue by getting a ticket and joining it – or by getting a ticket and promptly abandoning it. So information value is destroyed by the act of publishing it: telling people how long the queue is changes the length of the queue.1
In other words telling me that there are ‘54 customers in front’ actually tells me very little. It’s a production line view of the situation, not a user-focused view and it was a safe bet that there weren’t going to be 54 actual transaction before mine.
So I waited. And indeed the numbers rolled past faster than they should have done. I wasn’t counting, but at a rough guess, somewhere between a third and a half were no shows. And as I waited, three fairly basic lessons came to mind, applicable far beyond the post office
Modernising a service doesn’t necessarily make it better
Choosing an option on a touch screen which prints a numbered ticket has connotations of modernity which just standing in line in order of arrival altogether lacks.2 Being modern for the sake of doing things better is a very good thing. Being modern for the sake of appearing to be modern is not. Some aspects of the new way of doing things are undoubtedly better (people can sit while they are waiting, for example), but it’s not self-evident that the improvement is as overwhelming as the post office likes to think.
If there has to be a queue, the best thing to do is to make it visible
We would all prefer there not to be a queue. But if there has to be one, make it visible. That doesn’t have to be physically standing in order, smarter ways are possible too, as the picture shows.3 But whether it is a face to face interaction, a call centre or an order backlog, it is far better to know than not.
If you are going to give customers information, give them information which has some meaning
Telling me that there were 54 people ahead of me was at least an attempt to tell me something. But it suffers from the weakness that there is no easy way of converting what it is easy for the organisation to tell me to what I actually want to know. I wrote about two contrasting queues a few years back – one brilliantly managed, with visual queues supported by hard information about actual waiting times, and one which worked a bit differently:
If a customer had the temerity to ask how long they were going to have to wait, they were shown the pile of forms in the pending pile – which without knowledge of the number of staff on duty and the average time each one took to deal with, conveyed precisely no information whatsoever. Similarly, ‘we are very busy at the moment, please hold until you can’t bear the tinny music any longer’ is dramatically less useful than ‘we are currently answering calls in six minutes’.
With the data the post office ticketing system already has, it ought to be pretty straightforward to create reasonably accurate predicted waiting times. The fact that that is not the information they choose to provide is quite telling.
Meanwhile, today the BBC is reporting a much more subversive idea on queue management: serving the first person in the queue last and the last first. Let’s hope nobody from the post office gets to read it.
- Strictly this is a consequence of how the information is published: it is tangling it up with the act of deciding whether to join the queue which does the damage. ↩
- Or at least it’s supposed to. There is though nothing quite so dated as slightly tarnished novelty. ↩
- It was doing the rounds on social media a little while ago and the trail to the original source seems to be dead. I found it on Reddit, so it could have come from anywhere. ↩