Posts

10 July 2007

14 Comments

Airport express

There’s something about traveling that makes me think about interaction design. Maybe it’s all the signage and status updates and checkpoints. It feels like a physical representation of software (usually really bad software).

Recently I learned a few things about interaction design and user experience from waiting in lines at the airport. Nothing groundbreaking, but it’s interesting to confirm what you know by seeing it play out in another context.

A small increase in efficiency for all users has a more significant impact on overall performance than a large increase for a small percentage of users.

I never check my bags when I fly because of those “express” check-in kiosks. Almost every major airline has them now. I just get to the airport an hour before my flight, print my boarding pass and get in line for security.

With the big airlines, there is always a huge line to check in with bags. There’s never a line for the kiosks marked “no checked bags.” I blame all the security restrictions on what you can’t take on planes. I’d rather pack my shampoo in a plastic bag than stand in line for an extra half-hour. Most people don’t think about it that way I guess, so, shorter lines for me. Hooray.

Ten items or less

This is a classic example of the “express lane” method of increasing efficiency. I make a small sacrifice (packing light and carrying on) and in return I get to jump the line. It works great as a quick fix for a slow process, but only for a small percentage of users. For that small group of users, the increase in efficiency is very high, but it comes at the expense of all the other users of the system. The bigger the line for checked bags, the faster I get through the express lane. If everybody did the same thing, the lines would be just as long for everybody, and we’d all be back where we started.

But that’s the big boy airlines. This time I flew Southwest. The express lane trick doesn’t work with Southwest because everybody knows you just check in online and print your boarding pass at home the day before your flight. Because more people know they don’t have to wait in line to check in, more people choose not to wait in line to check their bags either.

So when I got to the airport, I blew past the check-in counters and headed straight for the security line. And I followed it all the way to where it ended, somewhere on the other side of baggage claim. I almost panicked. (“I don’t belong here!”) I looked around frantically for the express lane, but this was it. When everybody’s in the express lane, guess what? There is no more express lane.

Just as I started muttering to myself, the line started moving. And it kept moving. Before I had a chance to get over my inflated sense of entitlement, I was halfway across the baggage claim. I was through the whole line and putting my shoes back on at the other side in less than 20 minutes. It wasn’t exactly rocket fast, but it was steady. I don’t think I’ve ever seen an airport security line move that smoothly.

What impressed me most is that not only was the wait time completely acceptable (even if I had arrived less than 30 minutes before my departure time I could have made my flight), it was the same for everyone. This was a peak hour in peak travel season and nobody had to wait an unreasonable—or even uncomfortable—amount of time.

Express or excess?

So which is better, improving the whole process to get everybody through as quickly as possible, or a quick fix to get a few people through marginally faster while subjecting the rest to the same long waits? More importantly, which is more sustainable for a fast-growing system? What happens when you double, or triple the number of users? Which one breaks down first? Maybe the big airlines aren’t worried because they’re not growing users very fast—or at all.

Even supermarkets (the classic express lane environment) are learning the same thing, and are consolidating many lines into a single, faster-moving line. The results, according to this NY Times article about Whole Foods are shorter average wait times for everyone. The same principles come up in modern research on optimizing traffic. A recent
Berkeley study of San Francisco traffic” indicated that adding carpool lanes may actually contribute to congestion rather than reduce it.

These results are interesting because they go against how we naturally think about optimization—it’s counterintuitive. We see long lines and we think long waits. So we try to make the lines shorter, instead of making them move faster. But as is so often the case, it turns out our instincts are wrong because our intuitive mental model is flawed.

What are are the implications of this kind of thinking about optimization when applied to network systems like the Web? Should the FCC be asking the CEO of Southwest and Whole Foods about net neutrality?

What about our own apps? What can we learn about optimizing their performance, or even perceived performance? Which leads me to the other thing I noticed in the airport security line.

In order to get the maximum benefit from increased efficiency, it should be visible as early in the process as possible.

It struck me how hard it was for people (myself included) to get over the initial impression of seeing the ridiculously long line. Everybody had the same incredulous, panicked look on their face as they saw how far they had to go to get to the end.

Once we were in the line and realized how steadily and quickly it was moving, the panic subsided. But it took three or four airline employees to reassure people just joining the line and convince them to stay instead of milling around looking for other options. And even after telling everyone repeatedly that the average wait time was about 20 minutes, they still took a lot of abuse from passengers who had a hard time believing them.

No matter what you do to improve the efficiency of a process, you’re going to waste effort or lose users the longer it takes for users to see the benefit. People expect things to be slow, especially on the Internet. If you can convince your users up front that your process is going to be quick, simple and painless, the whole process will be that much more efficient. The perception of efficiency is almost as powerful as the real thing.

4:17 p.m.

Comments / Add yours…

  1. Jeff Croft 7:02 p.m. on Jul 10th, 2007

    Wow, has it been three months since your last post already?!

  2. Robin 8:29 p.m. on Jul 10th, 2007

    Yeah, that last point is great. Being able to 'feel' progress immediately is so important. I'm sure that if you put people in cars, for instance, and had them drive in traffic for thirty minutes from point A to point B, but in one case the traffic was slow & steady and in the other it was stop & go, the stop & go drivers would emerge WAY more frustrated, even though they'd gotten from A to B in the same amount of time.

    But here's a counter-example (sort of): At the Tampa airport (my fave airport in the whole world; it is by far the fastest & most efficient, even controlling for its smaller size) there is one central building with shuttles out to 'spoke' terminals. The shuttles are fast & frequent (they are going back and forth, not in a big loop) and you don't wait in any security lines 'til you get out to the spokes. So it's super-fast until you hit this little road-block at the end. But a) the lines are faster b/c traffic is distributed between all the spokes, and b) you feel like you're basically already where you need to be. You can usually see your specific gate just beyond the lines -- so it's hard to get TOO stressed out.

    Not sure there's any analogy for web stuff there. It's just a rad airport.

  3. Rob Goodlatte 9:21 p.m. on Jul 10th, 2007

    Airports might take a lesson from Disney World in line-waiting -- those folks have it nailed. They intentionally hide the line by making it wind through the attraction, and they put video screens, decoration, and theming along the length of the line to divert your attention from the slowness of the line. Airports could do the same by theming the line with interesting tidbits about the city you're or simply by placing TVs along the line.

    On the Web, I find breaking up forms into steps helps users tackle larger inputs. While I like to shy away from killer forms, if you present users with a manageable chunk of input and keep them informed about how much data they'll be adding, you get a much better experience.

    Same thing goes with DOM scripting, to a degree. If you can push stuff around client-side without having to wait for the server to keep up you make the whole process seem more responsive when in actuality it takes a long time.

    Fantastic post -- you really should write more often.

  4. Josh 10:03 p.m. on Jul 10th, 2007

    The military has been doing this for as long as I can remember when it comes to the lines at the commissary (the grocery store on base). You basically stand in one long line and someone at the front of the line tells you which checkout register to go to as they become available. This helps to maintain the efficiency of checkout by sending you to the next available register as opposed to you waiting in a randomly chosen line, only to find out that the person in front of you has double coupons on an item no longer tracked in the inventory...anyway, the system works.

    Though you wait in what seems to be a very long line, it actually turns out to be a fast moving one.

  5. Wilson Miner 11:51 p.m. on Jul 10th, 2007

    @Jeff: Shut up. ;)

    @Robin: The Kansas City airport is set up in a similar way and I love it. Each terminal is broken up into small groups of 3-5 gates, each with their own security checkpoint. The lines are always quick, and once you get through, you're right at your gate. It works great. I notice a lot of people complaining about it though ever since they stopped letting you bring food and drinks through security and all the restaurants and vendors are outside the security areas. All it means (since the lines are short) is that if you want to eat you have to do it before you go through security (they sell water and snacks inside the gate areas), but people get grumpy when you screw up their routine.

    @Rob: TVs would be nice, but they're still just a distraction from the wait which doesn't help if you're actually going to miss your flight if the line doesn't move faster. Also, I hate TV news and that's all they ever show in airports.

    @Josh: The military example is great because it illustrates how attached we are to our "freedom" of choice. We like the idea of being able to choose our lane in the grocery store and on the highway, when we're actually almost completely incapable of making the most efficient decision on our own (for ourselves, let alone for the whole system) even a small fraction of the time.

  6. Jeff Croft 12:23 p.m. on Jul 11th, 2007

    The KC airport is both the best and worst major metro airport I've ever been in. It's by far the fastest and most efficient to get in an out of, but it's also completly boring, and therefore about the worst imagineable plce to have a layover.

    But, no doubt, the layout does make for extreme efficiency, which I suppose is the most important thing.

  7. Sean Madden 3:56 p.m. on Jul 12th, 2007

    Please do not accuse me of missing the point, for I get the point and simply wish to slightly fork the discussion. One of the interesting things with the example you set is that the two systems in place (the auto-checkin and the security line) are run by two very different organizations with separate agendas.

    My guess is that the airlines are most interested in express kiosks and online checkin systems as a means to cut cost and reduce staff. A similar situation exists in supermarkets with their new-fangled self-checkout systems. Their whole purpose is efficiency and self-service, which has the mutual benefit of being both customer and organizationally friendly (at least so long as you don't take the perspective of the individual employee).

    What I don't know is who runs the security checkpoints and whether or not they are interested in efficiency. Does TSA run them or does the airport? If I work for TSA as a wander, I'm probably not all that interested in moving faster and pushing more people through. My day as a wander is going to be 8 hours regardless of how many people I get through. In fact, moving fast is more likely to cause me errors.

    Who answers to this problem? Efficiency as people pass through the security line is probably the number one concern for the traveller, but who is accountable for it at an organizational level and are is there any way to measure the efficiency and effectively balance it against security?

    The whole foods example is a very powerful example, but only because its a closed system entirely within their control. In a way, it's a luxury because they can afford to staff enough cashiers to make it fast. I am unaware if there is enough budget to staff security checkpoints at airports and achieve a similar gain.

  8. Andrew Dupont 9:32 p.m. on Jul 12th, 2007

    The particular aspect of this that interests me — the NYT article on Whole Foods touches on this point — is the need to overcome the customer's perception that longer lines are always worse. As a species, we're really bad at these sorts of estimations. Even though people realize that a consolidated line will move at a faster relative speed, we're still subconsciously trained to assume smaller lines will get us out the door more quickly.

    Because the benefits of queue management are obvious and significant, I think the goal is to minimize a person's reluctance to get at the end of a large line. As Rob points out, amusement parks do this, both by disguising the line's length at a ride's entranceway and by giving people feedback on how quickly the line moves ("20 minute wait from this point," etc.). I doubt a grocery store could employ the first technique, but the second is far easier.

  9. Dennis McClendon 5:42 p.m. on Jul 14th, 2007

    I don't know. Fry's Electronics uses both techniques: distracting you by diverting you through the magazines and candy, and having one line for many checkers. But it seems to me that huge amounts of cycle time are wasted in walking all the way from the throat of the line down to checker number 46.

    On the other hand, at the Arby's around the corner from my office, about a year ago the patrons spontaneously began forming a single line to feed the four registers (instead of four lines). The physical layout of the store seems to contribute to the perception that it's the right thing to do, and there are enough return users who know the drill that newbies seem to get the idea.

  10. alec a 7:25 p.m. on Jul 31st, 2007

    very interesting...... theoriests and professors study this for a lifetime in the mfging sector, they even write books on this stuff and come up with fancy names such as "Drum, Buffer, Rope". We study all the steps in the process and then determine which step causes the que to back up and then try and fix it. Once fixed, we study the processes again and find a different que is causing a new backup. Interesting at first but never solved only managed. I think Disney does the best by even looking at the pedestrian flow between attractions and perhaps detouring patrons to slow line growth. I wonder if they even back it up to the point before you even start to go to the park like the busses and trams. Now, don't go to India because they are the masters as hurry up & wait especially when the line has only 4 persons in it.

  11. Andy Budd 6:24 a.m. on Jan 4th, 2008

    A single big line normalises waiting times, making everybody's experience equal. You don't get into the situation where you're looking at the line next to you moving quicker and being annoyed that you chose badly.

    The question is, should everybody's experience be the same. If you're doing your weeks shopping you probably expect to be in the store a long time, so the length you wait is proportionate to the time you spend in the store.

    However if you're just popping in for a pint of milk or a paper, you're obviously in a rush. Is it fair that you're forced to wait much longer in relationship to the time you've spent in the store. The same length of time as somebody with a cart full of shopping?

    The same is true in airports. If you're a regular traveller and spend a lot of time in airports, you'll pack light and check-in online to minimise the time you spend waiting. This is because you seem to spent your whole time in airports and it's not fun. If it's your one big family trip of the year and you're ravelling with everything except the kitchen sink, you probably expect to take longer checking in, and you probably don't mind as it's a one off event.

    So while I agree that the system as a whole is more efficient if everybody has to wait the same length of time, I disagree that it improves overall satisfaction. Those people who were expecting to wait a long time save a bit of time, but in relation t their expectations they probably barely notice. Those people who were expecting a fast service now have to wait significantly longer for little gain, and are dramatically worse off.

  12. Wilson Miner 10:31 a.m. on Jan 4th, 2008

    All very true Andy. Obviously the holiday traveling family with 10 giant backs to check is going to spend extra time waiting in the line to check bags, and the single-carryon business traveler is going to head straight to the security line. So obviously the carryon traveler gets a boost there.

    But the security line is going to be long based on overall traveller volume (time of day, season, etc.) not anything directly in the passengers' control. So both benefit from speeding up the security process and making the effective time difference between big lines and smaller ones as minimal as possible.

    In my Southwest/Oakland example, I was impressed that I got through the biggest security line I've ever seen in about 20 minutes. I've waited in shorter lines at SFO that took almost an hour.

    Of course, after waiting in all ridiculous holiday airport lines, I'm convinced the entire airport security process is broken beyond hope anyway. :)

  13. Alistair Hutton 5:51 a.m. on Jan 16th, 2008

    The single queue model is pretty much standard in Britain. Bill Bryson has a piece describing his suprise in a (British) post office where there were two positions open and no barriers or other indicators for how people should queue, yet, the customers spontaneously formed into a single queue.

    Makes me proud to be British.

    Toodle pip.

  14. bliss 5:27 p.m. on Jan 23rd, 2008

    1. you should get a phd from northwestern university's transportation center. something tells me that you would have a high aptitude for that kind of stuff.

    2. southworst is a disgusting excuse for an airline, and p.s. all major airlines allow you to print your boarding pass at home 24 hours before the flight. also, are you not considering swa to be a "big airline"? it is.

    3. usually the reason the security lines get out of control is that people are retards and don't remove belts/coats/laptops/liquids no matter how many times they are instructed to do so. it's not so much a crappily designed system problem as it is user idiocy.

    4. if you ever want to know about the other side of the airline equation (the bags, the planes, the crews, the rasm/casm, the delays, and how all of that works) just holla. i am obsessed with air travel and i used to work for an airline, so i know...well...pretty much everything about it. also, my god mother was a flight attandant for 20 years so my spiritual guidance comes in the form of mile high club stories.

Post a comment