The (un)friendly skies: a quick look at Tweeting travelers
Holiday travel is always crazy. Most of us have a horror story about delays, cancellations, lost bags, rude agents, or worse. My colleague Masa—a Japanese national—has had his own joy dealing with travel, including circumstances that led to him working from Vancouver for a year. He probably finds that less amusing than I do.
Masa is an analytics whiz. He and I talk regularly about what we can do to push the envelope in research—methodologies, analyses, resources—to provide new ways of providing insights to our clients. Back in December, we decided we’d play around with collecting and analyzing tweets. Specifically, we thought it might be fun to see what people were saying at airports during this crazy period and cut the data a few ways to see what we came up with.
We set out to capture all tweets from DFW, OHR, and ATL (hubs of American, United, and Delta respectively) from December 22 through January 1. We then filtered down to the tweets that were about airlines, flights, or travel, and then coded them as positive, negative, or neutral and sorted by issue. Armed with this data, we looked at what passengers were saying about these three airlines. The results were interesting.
On the surface, it’s nothing surprising at all. Overall, there was a fairly reasonable balance between positive and negative. A recent study by Maritz Research, as reported in Marketing Research, an AMA publication suggests that web-based samples have a more negative bias than traditional surveys. Although this was not exactly the same kind of study Maritz looked at, one would logically conclude that the results in our exercise are also biased toward the negative, rather than a full collection of opinion. This measures up to common experience, too, which tells us that people complain publically more often than they compliment publically.
People complain about what you’d expect them to complain about, mostly delays and cancellations. Even the “thanks” in the world cloud below is frequently sarcastic.
But a few things pop out that wouldn’t come through a traditional survey (or, I wager, a traditional qualitative study like a focus group).
1)Treating people like people makes a difference. For all the complaints about rude reps, AA seems to have figured out that you can defuse some really vitriolic hate-tweeting, if you engage those people directly. Take one example below:
Partly, this also requires an airline—and I can only assume AA has done a better job with this than Delta or United—to publicize its “complaints department.” If you give people an outlet and engage them, you can lower the volume and the intensity of the animosity. If you don’t, they’re just going to shout a lot, and depending on who’s doing the shouting, that may have repercussions. (There’s an analogy here to Arab Spring, but we’ll save that for another article.)
2)People love telling you they’re in “the lounge.” Travelers—especially AA and Delta travelers—love talking about the lounges. Some of this is probably that they actually love the lounges. Some of it may be that they love making sure people know they are special and in the lounge. But if the latter were the main driver, we would expect to see more positive comments across the board (as we do with AA and Delta), not the negativity we see among United passengers toward the lounges. This suggests a couple things: 1) AA and Delta lounges offer a different and more-loved experience than United lounges (more/better free booze? more seating?); 2) none of the other major perks advertised by these airlines (first bag free, priority line, etc.) receive much public praise at all in the tweets, so at least with this crowd, having a good lounge experience is a good way to get some positive feedback.
In our experience in a variety of countries, for both political and corporate clients, having this 1 tangible thing to point to as a metric of “we are listening and improving” is really important.
3)There’s something in the water at ORD. It was interesting to see the slight differences in the way folks traveling through the different airports vented their anger. People in Chicago tended toward the more extreme “F-bomb” but shied away from the less aggressive S-word, much more than DFW and ATL travelers. I’m not even going to speculate as to why this is, and ORD definitely had worse weather during this period , but I’d be lying if I didn’t admit that Masa and I thought this was really interesting territory for additional research—something about “knowing your audience” and “meeting your customer where they are.” It reminds me of this ad campaign by Newcastle. Appropriate for New York, but maybe wouldn’t make as much sense in Seattle.
 To conduct this study, we first defined a circular area covering the entire airport. Then for each airport, we obtained the geographical center of the airport and estimated the radius (each airport has different size). Within this area, we collected all tweets with originating geocodes inside this circle. In total, we collected 16,579 tweets (6,441 from ATL4,362 from DFW, and 5,776 from OHR. Although we quantify data below, it not based on rigorous methodological standards of a scientific quantitative study.
 Maritz was looking at TripAdvisor vs. a traditional survey. The article—unfortunately behind a pay barrier—goes into detail about the differences and the methodologies in the studies. This article uses shorthand to describe just one result of their interesting study. If you have access to a copy, it’s worth a read.