Delta is Going to Start Overbooking More, and We Should All Be Thrilled

Delta

One of the big questions coming out of the pandemic is around what changes that occurred will remain permanent and what will shift back to “normal” eventually. On the revenue side, we have American believing that wide, sweeping changes have create a new normal. Delta has taken a more conservative and measured approach.

For that reason, I’m always interested in seeing what Delta says about this new world, and that means listening to President Glen Hauenstein give his opinions on the quarterly earnings calls. Last week’s call gave some interesting nuggets, and today I want to focus on Delta’s efforts to start overbooking more. This is a good thing. Get excited. Seriously.

In recent years, Delta has not been afraid to oversell its flights. It bumps more than twice as many people per enplanement than American or United.

Denied Boardings per Ten Thousand Enplaned Passengers by Marketing Airline FY 2022

Data via DOT ATCR, numbers are per ten thousand enplaned passengers

That being said, when Delta bumps, it has no trouble paying the price to get people to take a different flight on their own, or what the industry calls a “voluntary denied boarding.” You can’t see it on the chart, but during 2022, it reported a whopping 2 people that had to be involuntarily denied boarding, and those are the ones you want to avoid.

My assumption here is that Delta solicits earlier, more often, and is willing to go to a higher price in its bid to avoid bumping people involuntarily than American does. Like Delta, United almost never bumps someone, but it also presumably overbooks less since its voluntary rate is much lower than Delta’s.

So, it was with interest that I read Glen’s comments on needing to increase overbooking due to a higher rate of no-shows. The airline’s Q1 load factor was 81 percent. This was a huge increase vs Q1 2022 when Omricon was still wreaking havoc on travel demand, but it’s a couple points lower than where Delta had been in pre-pandemic Q1s.

Delta calls the percentage of people who show up the “materialization” rate. In other words, what percent of bookings actually materialize versus being canceled or changed? These days, behavior has changed and people are changing more frequently than before. You could say that this is due to the disappearance of change fees, creating less friction for people to change their flights. But the removal of change fees was also a response to changed customer behavior during the pandemic. So there is something of a “chicken or the egg” question here.

Either way, with more people not showing up for their booked flights, Delta wants to start overbooking more. But as Glen explains, the airline is approaching this cautiously.

…So, if you were [booked to] 103% [of capacity] on average and you have 2 extra points [to fill due to lower materialization], you just go to 105% in terms of what your ability to take is.

There’s a little bit of risk in that, and so we probably won’t go to 105% right away. We go to 104%, see how that works, 104.5%. And that’s why you have to retrain yourself and see what the actual events happen because these are changing in relatively condensed time periods. We don’t want to overshoot and cause a disruption, so we’re going to be a little bit more careful on getting that real-time.

The key here is finding consistency. I don’t think Glen was using real numbers here, but let’s just pretend he was. Let’s say you consistently have 5 percent of travelers not showing up. You would then book every flight to 105 percent, knowing that every single time you’d come in right on the nose.

But that is not how demand works. It’s choppy. I remember in the old days at America West, we had Friday night flights from LA to Vegas that were full of no-shows. My memory gets fuzzier every year, but I seem to recall we’d overbook those things by a third or more, and the airplanes would go out with empty seats. But try doing that to a flight from Omaha to Las Vegas and you’d have a very different profile.

Even that madness is solvable, however, if each route is consistent. The problem is, we’d have some Fridays where everyone would show up for those flights to Vegas. It was an unmitigated disaster. But you play the averages, and you also have to respond to the environment around you.

Over time, airlines have begun to overbook much less. Take a look at these eye-popping numbers from 1999.

Denied Boardings per Ten Thousand Enplaned Passengers by Operating Airline FY 1999

Data via DOT ATCR, numbers are per ten thousand enplaned passengers

These numbers look absolutely bonkers compared to today, but that’s because things changed. The feds jacked up the amount airlines had to pay for involuntarily denying someone boarding. Then, of course, Dr Dao got dragged off a United Express flight by the local police. That made airlines change their procedures to pay a lot more to avoid involuntary denied boardings. Costs skyrocketed, so airlines backed off on overbooking as much.

Now we live in a new world of no change fees, and behavior appears, we hope, to be solidifying into a pattern. With a pattern, airlines like Delta can start increasing overbooking rates to fill more seats. Glen’s comments are encouraging.

In case I wasn’t clear, we should all be happy about this. Having no change fees is a clear consumer benefit, because it allows people to change their plans more freely. That is not good for airlines unless it’s happening at a consistent rate. If it is consistent, they can just overbook more and offset the impact of having no change fees. Everyone wins.

If airlines can get a hold of this issue, then it bodes well for change fees remaining off the books in the future. I’m hopeful that’s what is happening here.

Get Cranky in Your Inbox!

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

26 comments on “Delta is Going to Start Overbooking More, and We Should All Be Thrilled

  1. Agreed, Brett. With fares still pretty high, anything that allows airlines to increase the supply of tickets sold (within reason) is a very good thing for pax.

    The two involuntary DBs are key for me. Even as DL does overbook more, and that number increases some, the point is that if you really, really have to (or want to) get somewhere, the odds of being involuntarily bumped on Delta will still be ASTRONOMICALLY low.

    1. The odds of being bumped period, let alone involuntarily bumped, on all airlines remains astronomically low. Those DB numbers are out of 10,000 passengers. Even AA only IDBs 0.0046% of passengers.

  2. What about Southwest? They have always had no change fees, so one might expect their pattern of last-minute changes and cancellations to be more consistent over the years, and relatively high. They claim that they don’t overbook, yet their flights are typically full. How does that happen?

    1. Southwest may not have change fees, but pax are still on the hook for the fare differences for any changes, and those can be pretty high at the last minute.

    2. From my experience, WN seemingly always has a large standby list. Every time I have tried to fly standby on WN, there will be 10-ish open seats and 8-10 standbys. My theory is that this might play a role in filling those no-show seats.

    3. Ron – Southwest is doing some kind of overbooking, at least the bumping numbers show that to be the case. I can’t quite figure out why their numbers are what they are.

  3. I guess I am wrong, but I would have thought the elimination of change fees would have decreased no shows. Pre Covid there was often no incentive for canceling as opposed to no showing.

    Are too many people just waiting to the last minute to cancel now?

  4. FWIW, I think the elimination of change fees was a mistake. I *also* think that the previous scheme, of charging a flat rate all the way from 24-hours post-booking up until the flight closed was a mistake too. A steep, $100’s, fee for flight that wouldn’t take place for several months was needlessly customer-hostile. And no fee at all for anyone that remembers to contact the airline before their flight also makes little sense.

    To me, a sensible structure would be to charge an escalating rate as departure date approaches, just as cruise lines do. Something like a nominal fee for up to 60 days out, going all the way up to 75-100% for 24-hours or less before departure.

    1. Didn’t AS do this shortly before Covid, was it free changes before the 60 day mark or something like tha?

  5. Not to mention that more VDBs mean more lucrative offers (often in the four digits on Delta) for the passengers who happen to be in a position to accept them.

    1. Exactly. Tough for people to have a bad taste in their mouths about Delta if they agree to voluntarily take a voucher worth hundreds of dollars or more in return for arriving at their destination a few hours later.

      Not an option I volunteer for (I don’t care much for airline credits/vouchers, and default to assuming that the fine print will make them tough to use), but a great option for those who are willing to take it.

      1. I can’t speak for Delta, but for us, vouchers on American and Southwest are super easy to use — as easy as a credit card payment. We’ve never had a problem using them to book flights. The only issue we’ve found is that vouchers have an expiration date and must be used for travel before that date. Not really a problem if you fly often. When arrival time isn’t critical for use, we volunteer to get bumped all the time. Free money!

  6. I feel like the airlines have way more data and the means to use it now than in 1999 in terms of specific routes or flights can be overbooked. With that as well as much more sophisticated revenue management, airlines have gotten better at getting flights to as close to 100% load factor and no more whenever possible.

  7. You do a far better job of addressing this topic than other people but you still miss that overbooking targets change throughout the life of a flight. The analyst that asked the question said it was a nerdy question and DL’s exec responded fairly well w/o getting too technical.
    Ensuring that a flight will go out full as it was before does not mean that the flight will be more overbooked on the day of departure. DL’s exec explained – as Jason captures above – that DL will increase the overbooking rate earlier in the life of the flight in order to make up for a lower percentage of pure business bookings, more leisure and “bleisure” and more no-shows.
    Modern revenue management systems learn from history and DL is simply it is trying to push the booking levels in order to help its system “learn” how to fill a flight under the current environment faster than if they system “learned” on its own.
    It is remarkable that DL has as low of an invol DB rate as it does (industry lowest) given that other carriers do not even intentionally overbook and yet still have far more invol DBs. It is equally remarkable how well DL’s revenue management has come given that DL was the IT laggard for years but now has clearly some of the best RM technology and processes – along w/ the airports – in order to lead the industry in both operational and financial metrics.

    1. There are a whole lot of inaccuracies here, so I’ll wade in and tackle a few:

      “does not mean that the flight will be more overbooked on the day of departure” – For an individual flight, sure, that’s true. For the system as a whole, it absolutely does. You’re not going to increase overbooking numbers and not see and increase in denied boardings. DL knows this, which is why they said they’ll do it in steps to see the effect. Changing those targets over the course of the flight will help, but there will be an increase.

      “Modern revenue management systems learn from history” – I state from extensive experience that this is categorically untrue. Most modern RM systems do no such thing. They rely on analyst learning to make those corrections. Now, to be fair, DL, as one of the world’s bigger airlines with a bigger budget, may indeed have such an RM system, but these systems as a whole are not machine learning-based and do not learn, on their own, from history.

      “given that other carriers do not even intentionally overbook” – What the heck? Of course they do. AA does, UA does, plenty of others. I think B6 announced they would stop, but they may have even restarted.

      “far more invol DBs” – Context matters. Sure, technically, AA has far more IDBs. But your chance of of IDB on AA are still only 0.0046%. So yeah, it’s less likely on DL than on any other airlines, but the chance is nearly identically 0 on all US carriers. It just doesn’t happen very often, period, on any airline.

      “in order to lead the industry in both operational and financial metrics.” – Um, did you check Q1 results? Have you checked their cancellation rate recently? The airline industry is very much a “what have you done for me lately” industry, and DL has not performed nearly well relative to peers the past little while. It remains to be seen, of course, which trends hold and which reverse.

      1. first, the DOT has released data for all of 2022 and for January 2023. Delta ended up with the best on-time of ALL US carriers in 2022 and Jan 2023 and the lowest cancellation rate of the big 4 in the same time periods.
        Everyone is waiting to see Delta fall from its position – but a few weekends or even a string of five days don’t define leadership or lead to loss of it.

        Revenue management systems DO forecast from history. Analysts CORRECT the forecast. That is all Delta is doing.

        Delta specifically said it is NOT their intention to increase day of departure overbookings – increasing the amount of bookings that are taken earlier could lead to more day of departure overbookings but they specifically said that is not their goal.

        and Delta still has a lower ratio of invol DBs. The fact that B6 and others do now overbook confirms that they have seen that DL can make overbooking work w/o causing passenger backlash. None have bested DL at the ratio of invol to vol DBs and the data shows that.

        1. I’m not going to argue your other points, because I made my points and I’ll leave it at that, but since this is my particular area of expertise, let me point out that these are NOT equivalent statements:

          1. Revenue management systems DO forecast from history
          2. Modern revenue management systems learn from history

          The first one, which you made in your response is correct. The second, which you made in your original post is incorrect. Forecasting is not learning. Forecasting is a simple statistical technique used in many industries, including the airlines. Learning is a far more involved processed that is just barely starting to make itself felt in aviation.

          Misquoting yourself to correct yourself and claiming you were right all along is wrong on so many levels and illustrates yet again why no one wants to interface with you. It’s either a lie or a falsehoods or a misdirection or moving goalposts or some other nonsense all the time.

  8. I’m not surpised to see Americans Involuntary Denied Boarding Numbers so much higher than Delta and United’s.

    We got Involuntary DBed on American back in January 2020 when the contracted gate agents in South Bend barely made any annoucements asking for volunteers to give up there seats (and at most mentioned a $200 voucher for taking a later flight). We were chosen because we were on Basic Economy tickets and weren’t agressive enough at joining the line for seat assignments.

    I filed a DOT complaint over the whole experience after the gate agents tried to force us with a voucher that less than a $100 more than the cash I knew we were entiled too. I knew my rights and demanded cash, but the agents ended up having to mail us checks (return address from the airport, not from corporate AA) to us that arrived within a week because none of the gate agents on duty had been trained on how to use their check printing machine. There was one other gentleman who was also Involuntarily Denied Boarding on the flight who was very greatful I had informed him he was entitled to cash instead of a voucher, after he was about to accept a voucher since the agents violated DOT rules and didn’t properly inform us of our rights.

    During the whole ordeal as we chatted with the contracted agents they told us they also handeled United flights and said United’s computer system actually gave them the power to offer higher voucher amounts to get volunteers and they felt constrained by American’s computer system that didn’t really empower them creating our situation.

    1. ” none of the gate agents on duty had been trained on how to use their check printing machine”

      This line cracked me up.

  9. As a retired DL employee this scares the hell out of me. It’s hard enough to use a benefit, that I worked 31 years to have. With more overbooking, which will mean higher load factors, I might NEVER get on a flight using this “benefit”.

    1. It should scare you, the benefit is gone. A retired airline friend (AC) just bought her way to Barcelona to catch a cruise. She figured going standby was a lost cause.

    2. An 81% load factor means that there are still many flights going out with empty seats. It’s just the most popular flights at popular dates and times that are full and need VDB/IDBs. If you’re flexible with times and don’t try to go to the most popular places, standby can work just fine (which is also true to some extent for better deals on paid tickets)

  10. There has also been significant improvement in forecasting models from 1999 to today. I have to imagine airlines have improved their predictions of no-show rates by flight and that is a contributor to lower bump rates.

  11. I hope DL is,not trying to do this too much on International flights. When the next flight is the next day, being “bumped” or routed a very long way around is a pain. I also suspect that no shows are fewer on ainternational flights

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