Southwest’s Surprising Decision to Stop Overbooking


If you thought United’s 10-point plan to deal with the Dr Dao dragging incident was a bit much, just look at what Southwest has done. In what looked like a surprise announcement, CEO Gary Kelly said on CNBC that Southwest would stop overbooking. Later comments during the airline’s earnings call indicated this was in the works before, but the timing was directly related to the United incident. Everything I know says this is a bad idea, but I spoke with Andrew Watterson, Chief Revenue Officer at Southwest, and he says otherwise.

I’ve defended overbooking before, but for an airline like Southwest, it should be even more important than for others. Overbooking is a great way to compensate for the fact that people don’t always take the flight they book. Leisure travelers? More often than not they do. And that’s why when JetBlue was starting up with entirely non-refundable fares and a highly leisure-focused crowd, it might have seemed like a more minimal risk to get the PR benefit of not overbooking. But Southwest is quite different.

Southwest may have started as a low cost carrier, but it was always a business-focused airline. Remember, when it started, it put a whole bunch of frequencies into the Texas triangle (Dallas-Houston-San Antonio). Having high frequency encourages people to make changes since there are so many options. Meeting finish early? Grab an earlier flight. With a high percentage of people flying on a refundable fare at Southwest, there is often no cost to making that change. And even for those on discounted fares, there’s no change fee; just the fare difference applies. A recent policy change now allows A-List elite members on those cheapie fares to standby with no charge. With high frequencies and low fees, Southwest is an airline built for people to switch off flights at the last minute. And that means overbooking should be important for filling seats and keeping fares low.

What Andrew explained, however, is that those high frequency routes where that behavior is most prevalent, like in Texas or from LA to the Bay Area, don’t usually have a need to overbook anyway. Those routes tend to have lower load factors, so in most cases, overbooking wouldn’t add much to the bottom line. It’s the less frequent, longer routes like, say, Caribbean flying or Vegas/Florida stuff that are going to be prime targets. But those are full of leisure travelers so the rate of people not showing up or switching at the last minute is minimal.

As someone who’s been in the industry a long time, Andrew empathized with my disbelief at what he was saying. In fact, he said that even a couple of years ago he wouldn’t have believed this would be the right path. But in the last 5 years alone, he says the revenue benefit from overbooking at Southwest has been cut in half. That’s a dollar amount, not a percentage, so the decrease is magnified when you consider Southwest’s growth in the last 5 years. Once they looked at the numbers, they balanced that against the operational issues with handling overbooking situations and decided it wasn’t worth it. Working an oversold flight is time-consuming for a gate agent, and to be fair, Southwest isn’t very good at overbooking so it makes for a higher burden at the gate.

Wait, Southwest isn’t good at this? Last year Southwest bumped more people involuntarily than Alaska, American, Delta, and United combined. Its rate of having basically 1 person involuntarily bumped for every 10,000 flying was the second worst in the industry behind ExpressJet. That’s not good, but let’s keep this in perspective. We’re still talking about maybe 40 people a day, and I’d bet in most cases there were decent reaccommodation options.

The big open question for me, however, is whether Southwest could simply do a better job of overbooking, reducing the operational burden and providing a greater revenue benefit. With its old reservation system being put out to pasture next week, Southwest will have a shiny new system that should provide better data and help the airline make more accurate forecasts. To me, this seems like the right way forward, but Southwest still doesn’t think there’s enough there. The airline would rather simplify its operation. No question it also gets brownie points in Washington for this since Congress is starting to stir on the issue. In fact, this move could have added fuel to that fire, making life harder for the big 3 if it gets any traction.

This, of course, doesn’t mean Southwest will never bump another person. It will need to substitute smaller airplanes, block broken seats, and get saddled with weight restrictions. When that happens, people will get bumped. Just ask JetBlue which has a rate of involuntary denied boarding that’s not much better than Southwest despite the fact that it doesn’t overbook. Still, for Southwest, the number of instances where this is an issue will drop significantly. And yes, that little PR bump right now can’t hurt.

This does mean that Southwest will be selling fewer tickets for the same number of seats. The revenue benefit may not have been huge, but it’s still a benefit nonetheless. The airline has no 1st or 2nd checked bag fees, no change fees, and now no overbooking. That’s a lot that has to get baked into the fare.

Everything I know tells me this is a bad idea, and my version of this post last week was a lot harsher. But I also know Andrew knows what he’s doing, and he felt the same way until he looked at the data. In this case, he says Southwest is letting the data drive the decision, and the data says it’s ok to stop overbooking. If you need me, I’ll be in the corner in the fetal position trying to make sense of the world.

Get Cranky in Your Inbox!

The airline industry moves fast. Sign up and get every Cranky post in your inbox for free.

38 comments on “Southwest’s Surprising Decision to Stop Overbooking

  1. Brett – I’m as surprised by this as you are. I also believe Southwest (and its consultants) analyzed the data on this thoroughly, and although data are always subject to interpretation, I have no reason to doubt the stated logic. (Reminder – I am a former Southwest Employee)

    One clarification: I don’t know if it’s still true, but one reason Southwest historically had a surprisingly high rate of involuntary denied boarding was that DOT included in that calculation passengers who were 10-minute ruled. With its high frequencies in short haul markets, just as passengers were able to catch an earlier flight, so too were they able not to worry about showing up so close to departure that their seats had been given to stand by passengers. This always accounted for Southwest’s seemingly poor performance on this metric, and it was never really fair to Southwest.


  2. So this works for WN because the routes where people change flights often have low load factors??!! Why is it that every DL or AA flight I get on is 100% full? Maybe that’s why WN costs more and requires a connection at Midway. Honestly it sounds like they are running ops from the 1980’s. I’d be none too pleased to hear this if I were a Southwest investor.

    1. full flights don’t mean anything about yield. Delta and AA may be more “full” with non-revenue standbys, or with sufficiently cheap fares that the marginal revenue of the overbook isnt much over the cost to deny someone else + operational impact of stress, delay, and unhappy customer perception.

      DL in particular, has many more fleet types and can better match size and demand. WN can’t do that, but also doesnt need spare crews, training, and parts for 10+ different aircraft types.

  3. Why not sell a provisional fare when the capacity is reached? Lower price, we think you’ll get on but we can’t guarantee it, and we promise to roll you over to the next available flight if necessary. You get some added revenue and you tell the truth. Our industry (I’m retired now) has a long history of duplicitous behavior toward customers. Are we ever going to get it? Tell the damn truth.

    1. that introduces much more complexity, not to mention how to distribute it across OTAs unless you designate another fare class. You can typically tell though – Y class economy tickets with no seats on the seatmap. While not 100% accurate, it is a very strong indicator.

      Also, the advantage to overbooking is to charge someone more than it costs to displace someone else (or at least fraction of expected passengers). If you sell it at a discount, the operational impact isn’t worth the benefit. Might as well not sell one of the cheapest tickets to begin with.

  4. > Those routes tend to have lower load factors, so in most cases, overbooking wouldn’t add much to the bottom line.

    I’m not understanding this early on a Monday morning. Can you please explain it like I’m five? If a flight is only, say, 60% full, then why wouldn’t overbooking help? Is it just that with flights that are not very full, there isn’t much demand, and therefore not many opportunities to truly sell more seats than the plane has?

    Also, I worked with Andrew in a previous life, and remember him as a very sharp and fun-loving guy. If Andrew says that there is little benefit from overbooking, I will be more than happy to take his word for it.

    1. Kilroy – If the flights aren’t full, then there’s no way to overbook. It’s only if a flight is full that you can do it.

      1. On the contrary. If you have multiple flights that are booked to say 80% and one of those flights cancels, you have just booked up several flights that weren’t previously full much less overbooked. Now you have several flights that have to be processed as possible oversells when previously you had none.

  5. Count me in as a skeptic. Sorry, WN does everything possible to eek an extra micron of productivity out of its people, planes and real estate, Even if the dollar value of overbooking has been cut in half, that is still some $ benefit. And from what I understand, with them no longer really being a LCC, they need every penny of $ they can get. I guess it’s just inconsistent with everything else they do…

  6. I wonder, with their new booking system coming out, the Wanna Get Away fares will become truly non-refundable, and no credit given? It’s true that they have extremely flexible fare rules; when I was cancelling a ticket last minute a couple years back the agent even told me I’d have a year to use the credit (I was unfamiliar with this policy at the time). Hoping this isn’t the case, but I can also see them going this route!

  7. WN is simply managing to the public perception that denied boardings are bad when there are clearly big differences between how effectively airlines manage the process including that WN has the worst involuntary denied boarding ratio. They will leave revenue on the table in the process of trying to be “simple”
    Given that B6 and WN had far worse involuntary denied boarding ratios than the legacy airlines (including AS which uses legacy type revenue management systems), the problem is that airlines that want to convey the idea of simple, pleasant travel don’t have the tools or processes that the legacy airlines have.
    And most significantly, the fact that B6 has a higher involuntary DB ratio than the legacies despite not overbooking on a pre-departure basis says that no airline can eliminate the possibility of denied boardings because of operational issues and of course crew movements which drove the UA fiasco. If a flight is fully booked before departure you can bet WN isn’t going to tell its passengers at the next city that they can’t get the crew there because the incoming flight is booked full and WN doesn’t overbook. Airlines that say they don’t overbook make it far harder when they really do need seats. This is one example where trying to act simply in the name of a pleasant experience will backfire for low cost carriers compared to legacy carriers who live with the complexity that is part of airline operations, including changing usable aircraft capacities.

  8. Nobody ever talks about the millions that go every year in maintaining and improving the sophisticated analytics models and engines that handle overbooking, plus the cost of analyst and management to review it on a daily basis, plus the complexity added to the business, plus the cost to the brand every time you leave someone behind who didn’t want to be.

    Many projects add to revenue, but many do so by decreasing profits. Southwest found out long ago that bag fees are in this category, and it’s more likely than not that overbooking also belong to the revenue “enhancing” but profit lowering bucket. Good riddance.

  9. As a SWA retiree, I am going to trust that the company knows what it is doing and has made the right decision on this. 42 plus years of profits speaks volumes about the smarts at SWA. Gotta LUV em!

    All Day Ray

  10. I can see how Southwest could take this position. I realize Southwest’s “strategy” has changed significantly over the years, but network type also has a lot to do with it. While “voluntary” no-shows are one component of the overbooking factor, misconnects are another critical component.

    Back in the days when I used to work for UA, I was trying to take an A320 to BOS. The thing was oversold by like 50. I got on the plane anyway. Were there really 50 people who “decided” they weren’t going to BOS that day and canceled their reservation? Doubt it. But there were t-storms at the hub that day…

    I think the other thing that mattes is that WN’s “no change fee” policy actually encourages people to actually call and switch their reservation. With legacy carrier change fees being what they are, there’s less incentive for me to call and cancel my reservation if my plans change. This is particularly true when a new booking costs less than the change fee, which does happen.

    So, if there’s little reason to no-show at Southwest, and their network type leads to fewer misconnects, I can understand how they may feel overbooking isn’t worth the trouble.

    1. Some of WN’s “hubs” connect as much traffic as legacy carrier hubs. MDW and BWI are just two that are both heavily dependent on connecting traffic.

      Further, some legacy carrier revenue management systems are capable of distinguishing between misconnected and no-show passengers specifically because it affects the overbooking forecast.

      Technology exists to match any type of airline operation.

      And none of it matters in cases such as the UA case where the airline’s pretty accurate pre-departure overbooking system was overridden by crew members that “had” to get there and weren’t previously booked.

      What is missing in all of these “we don’t or won’t overbook” pronouncements is a recognition of the reality that there will be times when seats will be needed for crew or operational reasons that require that inventory be reduced below the number of booked passengers.
      WN’s policy change doesn’t address that including its statement that 20% of its involuntary denied boardings came from crew movements.

      If CF is right and WN DBs about 40 passengers per day – which is a small number of passengers compared to the number of passengers WN carries – it still could require that up to 10 passengers be involuntarily denied in an environment where passengers don’t have expectations that they should be denied boarding on WN.

        1. I understand… and the percentage of those that are crew related won’t be reduced. If WN still seats out flights hours before departure and then has to move crew, they will have to bump passengers.

          I agree with the thesis of your article and that there will be a revenue impact. The airlines that are trying to distance themselves from overbooking will hurt themselves financially.

          Overbooking, if properly managed is a good business process. Trying to argue that you don’t overbook and then having to oversell for very valid business reasons including weight restrictions, aircraft changes or crew movements only creates hostility with passengers.

          as for the $10k overbooking “standard” that has now been established, it is only a standard for those airlines that don’t have the technology and processes to find passengers who are willing to accept a denied boarding for far less – and those passengers most certainly exist

      1. I’m aware that WN’s network has morphed over time to deal with more connecting traffic. I’d have to crunch the numbers to see how their traffic mix compares with the legacies, but I’d wager that in any high frequency market, the mix of passengers leans toward O&D traffic over connecting traffic. I’d make this claim regardless of carrier — the frequency legacy markets are certainly not targeting connecting leisure traffic. Those intra-CA short hauls are such an example.

        I do think your last statement is creating some hairs to intentionally split. Every passenger (me included) buys a ticket with the expectation that they will be granted boarding to the aircraft and transported to the final destination by the time published in the schedule.

        While I have some expectation that I won’t always get there “on time”, I do expect the flight to operate (non get cancelled) and I do expect to get on it if I have a confirmed seat. Cancellation rates are actually quite low — in 2016, US airlines operated more than 98.5% of their flights, and in 2015, boarded more than 99.9% of the passengers.

        What I’m saying is that from a quantitative standpoint, I expect to be boarded on the flight for which I’ve purchased a ticket, but have less of an expectation that I will get there on time. Such a low rate of passengers are denied boarding, that switching over to a system that doesn’t overbook will not change my perception at all as to whether I’ll get on the plane.

        1. Dan,
          using DOT data, WN at BWI carries about 45% connecting passengers. ATL and MDW are both around 40%. There are legacy carrier hubs that have lower percentages of connecting traffic.
          Intra-California markets are largely geared to local passengers but markets like LGA/EWR/DCA-ORD/ATL/DFW etc are all high frequency markets where there are still high percentages of connecting traffic. Carriers use the high demand in in the NE cities to maintain high frequencies to their hubs which also increases their ability to compete in the local market.
          I agree with you about the expectation of getting there… but that is the whole nature of voluntary denied boarding compensation. There are passengers who are willing to change their plans for a certain amount of compensation – which obviously differs from person to person and situation to situation. Turning the cost of one airplane ticket into a 2nd ticket (or more) can be very appealing to some people.
          I am not sure what you believe I am splitting hairs over. Can you elaborate?

  11. At $10000 the ticket price (in the public eye) for bumping oassengers from flights, I can understand the mantra to not overbook !!!…

  12. The comment that Southwest’s longer routes are much less business-heavy struck me. I guess it’s not surprising: on DAL-SAT, business travelers don’t care about the lack of first class upgrades and other perks that WN doesn’t offer, whereas WN’s product is somewhat less appealing relative to the competition for frequent business travelers on longer flights.

    1. DMK – I knew someone was going to catch that! I’ll fully admit it was simply lazy photoshopping. I ran out of time and had to pick something, so I went with what was easy.

  13. I have to disagree about this overbooking thing. It is simply another revenue generator. The airline already has money from the first traveler and now it is making money from a second traveler by selling them something that does not exist (at least not yet). It is nothing but an auction scam whereby the airline tries to sell something they already sold you for more than you paid for it. In turn, if that works out for them, they then return some of that profit by paying you to get off the plane. It is, however, in the aggregate, less than what they received. If it wasn’t, they wouldn’t do it.

    I’m also not convinced by this “it is a minuscule” amount of money. If it really were minuscule then stopping it would have only a minuscule impact on revenues. Those who then argue ticket prices have to go up to cover the lost revenue from overbooking are, if the above is true, only talking about a minuscule amount of increase. But the fact is that it is highly profitable and is virtually all no risk at all.

    In my opinion, the very minimum that the airline should have to pay is the change fee they charge you to change a ticket…around $175 per ticket I believe. Then they should have to pay the difference in fees that they received in selling “your seat to someone else.” I think this should be legislated out of existence. It amounts to fraud in my humble opinion.

  14. Dreaming as I am, there will come a time when overbooking will come to a sad, hard to swallow, end. That is when, airline tickets can be freely bought and sold on the open market. You buy a ticket, for one of the 137 tickets on QuickAir flight 777, Monday, May 1 and it’s yours to do with it as you wish, security rules, of course. Airlines have become so smart with how they can overprice everything with capacity controls, etc., allocate fares to a million buckets, why can’t they figure out how to mange the sale of 137 tickets for each flight and go away fat and happy. Sure, lots of people will try to corner the market for that great flight to SuperBowl-land. Well how about every other seat event? Let the market do its job, what the traffic will bear!

    1. Jaybru – I’ve talked about this before, but the problem is connections.
      Sure, it’s easy to just sell each seat on an airplane at one fare. But then what if someone has to connect? You end up charging a lot more than you would today because it would have to be the sum of the two local fares. And then what if there’s a schedule change? Aircraft swap? There are a million issues that don’t happen in a stadium or arena where the seat count is fixed. I’ve found myself staring at the ceiling trying to solve this problem, but I’ve yet to figure out a good way to handle it.

  15. Southwest Airlines decision to cease over booking flights will be such a great bonus for employees, Customer Service Agents and Operations Agents and Flight Attendants  especially. It will offer a more complete positive customer service presentation to their passengers. It will also eliminate some departure delays which are directly caused by an oversell situation and the related costs associated with flights that run late.

  16. I think it’s ironic that overselling has been such a huge issue when it wasn’t the reason why that guy got dragged off the plane as has been discussed.

    However, I’ve taken a few bumps over the years and I’ve been happy with the compensation. Never thought of it as a huge inconvenience. If it was, I would have never taken the compensation.

  17. Gee, a business or corporation does something ethical and moral and you guys disagree with the decision putting business at the forefront and not people. How pathetic. It’s no wonder your country is in the state it is in, especially how you elected Denise the Menace.

  18. Gee, you guys disagree a decision when for a change a business or corporation finally does something ethical and moral, putting people first. How pathetic. It’s no wonder your country is in the situation it is where you elected Dennis the Menace.

  19. Most readers of this blog are probably too young to remember the old “adult standby” fares between LAX and ORD on night flights during the early seventies. These passengers were seated in the last coach cabin of the 747s I worked. The reason they were segregated was because they were to be served no snacks. This fare was created in order to lure more passengers onto lightly booked flights in a very competitive market pair during a time when aircraft available seats far exceeded demand. If I remember correctly, the ow fare was $88.00.

    WN may consider offering something similar on itineraries without connections. Many nonrevs, myself included, decide at the last minute to standby for a flight (was unable to get on several last month–no big deal). With the right advertising, I believe WN could create a culture of proud “adult standbys” willing to take a chance if the fare was attractive.

  20. Well, this would allow a cut back on the number of gate agents and removal of PR flash points. And what about removing the luggage of those bumped from the flights? There’s probably enough savings in the weeds to offset lost revenues, let alone the cost of resolving any public incidents that could occur.

  21. Cranky, I’m thinking the whole thing is a crock. It appears Southwest at some point is headed back to standby lists.

    In this case, the airline would take a passenger’s money, for a standby reservation on a full flight. As confirmed passengers fall off, the standbys are accommodated. If the passenger is not accommodated, the airline either rolls them to the next flight or refunds the fare in a future travel certificate.

    The airline would use frequent flyer status and fare size to prioritize the stand-by list. The airline gets to where it is today but doesn’t guarantee a seat and doesn’t have any ugly Dr. Dao incidents. The programming geeks could make this work in a day or two.

  22. As a former 737 Capt I agree with SWA decision. The United issue was handled badly. The Pax was wrong but the bad PR UAL got cannot ever be un done.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Cranky Flier