Queues at Heathrow: a numbers game
‘Heathrow passport control misses target’
Financial Times, May 3
How bad are the queues, actually?
Damian Green, the immigration minister, has told parliament that the worst queueing experience was an hour and a half, at Heathrow last week.
Sorry – the worst queue at Heathrow last week was an hour and a half, or Heathrow was the place with the worst queue, ever?
Good question. Meanwhile, Heathrow’s operator, BAA, released data showing the worst queue was three hours.
I’m confused.
The way in which the statistics are gathered and reported doesn’t help either. The Border Force is supposed to ensure that passengers from outside the EU get through immigration checks within 45 minutes 19 times out of 20, while EU-based passengers should get through within 25 minutes, again 19 times out of 20.
They aren’t exactly stretch targets, are they?
No. The Home Office has been proudly pointing to its latest report that shows all these targets have been hit with room to spare.
Hurrah!
Which tells us that things were going smoothly in October to December last year.
Hurroo . . .
Quite so. And normally we’d just have to wait for the Home Office to get around to publishing data for January to March and finally, in the summer some time, we’d get to hear about what happened in April.
So how come everyone is reporting that the targets were missed in April?
This is unofficial data released by BAA.
Are we comparing apples and oranges, then?
Not really. Both BAA and the Home Office use the same basic methodology. At regular intervals they pick somebody joining the back of the queue and then time how long it takes for that person to clear immigration. There is one difference: BAA picks a person to track every 15 minutes; the Home Office only does it once an hour.
So that might explain why Mr Green and BAA were reporting different numbers.
Yes. Also, the worst cases reported by BAA actually happened shortly after Mr Green had made his statement. Mr Green, incidentally, was trying to have his cake and eat it by using unpublished data to bolster his argument in the House of Commons. His numbers are supposed to be either official statistics or he is supposed to keep them to himself. I suspect the UK’s statistics watchdog will take a dim view of that.
Can we start talking about how to fix the problem now?
Patience. There is one nerdy point about the numbers worth attention. Imagine Heathrow has 10 hours in which 1,000 passengers an hour arrive and walk right through, followed by one hellish hour in which 10,000 passengers arrive and end up camping out half the night.
It’s easy enough to imagine – go on.
Well, in that case half the passengers have had a terrible time. But the way passengers are sampled by both BAA and the Home Office, it was only in one hour out of 11 that things went badly and so only one passenger out of 11 suffered excessive queues. The sampling method they’re using systematically under-samples times when the airport is very busy, which is the very time that queues are longest – Mr Green did, after all, blame “bunching of arrivals” for recent problems.
I get the point. So things may be worse than the official numbers suggest.
Yes. And since the official numbers suggest that almost a quarter of non-EU arrivals at Terminal 5 in April had to wait more than 45 minutes, the baseline is hardly great.
How do we solve the problem?
More staff. As a rough reckoner, if a queue is an hour long and you open up a new desk, every person you pull out of that queue saves an hour of queueing time. If an immigration official can check 60 people an hour then she is saving 60 hours of passenger time for one hour of her time.
But who is going to pay them? The government? The airlines?
The point is, the benefits are greater than the costs, so if the government can’t figure out how to make the queues go away they’re not fit for office.
At least the Border Agency is recruiting, right?
No, they’re going to cut staff by 18 per cent over the next three years.
This is insane! I’m going to complain to the Home Office.
Really? Join the queue.
Also published at ft.com.
8 Comments
Gina says:
Glad to see someone is treating Passengers as customers… I do think sampling should more accurately reflect passenger experience – what use is a count when there are few flights – although when there are severe delays, on account of restricted flights passengers could well experience horrible customer service across the board… Thanks for sharing
5th of May, 2012Tom says:
In the hour after that in which 10,000 arrived there will still be 9,000 in the queue, and 8,000 in the hour after that; or maybe fewer if Mrs May puts on more inspectors.
5th of May, 2012Prof Shravan says:
Absolutely adore your SIMPLISTIC Q & A style.
5th of May, 2012Tim says:
Both Heathrow and Gatwick have form on their own statistics for queuing time :
http://www.caa.co.uk/application.aspx?catid=14&pagetype=65&appid=7&newstype=n&mode=detail&nid=2046
And when does the queuing time start ?
There are anecdotes that people are being prevented from entering the Immigration Hall. So if they aren’t in the queue, the clock hasn’t started ? Yes ?
The Kellogg Institute blog has a couple of interesting articles on the problem and a nice chart of how times vary throughout the day on the other side of the pond :
http://operationsroom.wordpress.com/2012/05/05/customs-queues/
It would be interesting to see that info for Heathrow.
5th of May, 2012Gerard McGlew says:
“The sampling method they’re using systematically under-samples times …”
It is always amusing to hear of these attempts to beautify the numbers, with minimal thought to passenger experience.
Though there could be better metrics (as referred to in the Kellogg blog), they would be harder to collect. For the given metric, the most interesting fact is not the average, but rather the maximum length of time one of these end of queue passengers has had to wait. Who cares if the wait time at 3am is non-existent, if most people arrive at 3pm? Even a weighted average would be better: number of people in queue multiplied by wait time of the person currently sampled, and then averaged per person.
The idea of having pre-queues, but only counting once people have joined the correct queue has also been tried before, in other public services.
How about this for a metric: plane touchdown until train leaving time (or car reaches national motorway speed limit)? Average and maximum are both of interest.
6th of May, 2012Bevan says:
In the interests of a concrete suggestion, either of the following would provide a better metric:
1) Sample every Nth person’s experience (as long as N isn’t too large).
2) Sample every M minutes, but WEIGHT THE SAMPLE by the total number of passengers that passed through in that time period.
Neither is technically difficult, and #2 could probably even be back estimated using the existing sampling combined with gross-passenger numbers.
7th of May, 2012Matthew Taylor says:
Great article Tim, but for the government there is no cost associated with the 60 passenger hours queued, where there is a cost for the one extra person they would have to employ.
Would also be interesting to understand the methods the agency uses to predict the arrival of passengers. Contact centres have been modelling call arrival using ERLANGS for years, I wouldn’t be at all surprised if other operations still rely on passengers per hour processed rather than treating the bunching and unpredictability as something tangible. Maybe you could write an article on Mr Erlang and how everyday his calculations could be?
14th of May, 2012David Weston says:
The UK Border Force targets are clearly too lax.
16th of May, 2012On the busiest day for Heathrow arrivals (26 July), 138,000 are expected: if 1 in 20 are waiting for longer than the target times, that would be 6,900 people waiting.