United trounces Southwest in 2009 on-time performance
No, that's not a typo. When you look at the on-time performance statistics published by the DOT for 2009, you do see Southwest solidly in 2nd place with a 83.0% on-time rating. United is in 6th place, with 81.0%.
But how is that a relevant comparison? After all, Southwest flies to a completely different mix of airports than does United. Why should a head-to-head comparison take into account things like weather and Air Traffic Control issues completely out of control of the airline?
United itself started the fight when they began advertising this: United. #1 in on-time arrivals.
They felt they deserved this recognition because they beat the other network carriers: Delta/NW, US Airways, American, and Continental.
Some critics (me included) thought this was a bit bogus, since don't they also compete with carriers like Southwest?
Brett Snyder of BNET (and Cranky Flier) agreed, and wrote it up on monday: United Most On-Time for 2009: Count Us Skeptical
How does United get to claim credit for being "#1" when they were, in fact, #6? From the customer perspective, the competitive set is any airline who is competing for your business. In that case, while Hawaiian (The perennial #1) might not be as relevant, Southwest and Alaska (#2 and #3 in the rankings) certainly are. But the customer is making the decision about flying an airline for a given route -- they don't necessarily care if United's flights to Chicago are always late if they're flying from Denver to San Francisco, right?
What if we looked at the on-time percentage of United vs. Southwest on only the routes they fly in common. Would that not help us see whose on-time performance really was superior? Similarly, we can compare United to Alaska on routes where they compete.
The results: Skepticism proved unfounded!
For all of 2009, on only routes shared between United and Southwest (~68, give or take a few depending on the month), on-time performance was:
United: 81.7%
Southwest: 77.7%
Looking at it month by month, United beat Southwest in every single month of 2009. United truly kicked Southwest's butt in on-time performance.
The tune changes when you compared United to Alaska. On-time performance on like routes (~15 give or take):
United: 79.2%
Alaska: 80.0%
But the difference is minimal. In fact, they each beat the other in 6 out of the 12 months, with Alaska taking the overall by less than 1 point.
So what does this all mean?
1) United CAN be proud of their operation for 2009. They did an exemplary job, coming close to tying one of the best performers in the industry (Alaska), and soundly beating their low-cost rival known for its on-time performance on routes where they overlap (mostly out of Denver, a hotly contested hub).
2) Like-route analysis is a far more insightful way of assessing head-to-head on-time performance between airlines. Southwest flies to lots of secondary, sunny weather airports -- hence its overall rank of #2 for 2009. But when you compare the routes it shared with United (where they face similar operational challenges relative to airport, airspace, and weather issues), the story is completely different.
There are other critiques of United's claim as well (well outlined by Brett). One being that the data doesn't include regional carriers flying under the United Express brand. Although that is likely not a big deal for this analysis since the routes United shares with Southwest are served mostly by mainline aircraft. Additionally, United's biggest regional carrier, SkyWest, finished AHEAD of it in 2009 rankings at #5 (we're not talking Atlantic Southeast or Comair here, which surely would make Delta's performance far worse than even it looks at #12).
So my apologies, United, for ever doubting you. Your press release may have been overzealous -- but the spirit was dead-on. Congrats for a fantastic 2009.
From my end -- more head-to-head analysis to come. I've long believed that the way on-time performance data is summarized for public consumption leaves a lot of insights hidden (and promotes false conclusions, as seen above). I'll continue to try to pull out interesting, relevant, and actionable insights from the 2009 data set in the coming weeks. Let me know if you have any requests of head-to-head airline comparisons or other interesting analysis.
Evan
Posted by Evan
