Bloomberg published an article yesterday about a little test United is said to be running that will pay people to give up a seat on a flight several days in advance. If you just read the top of the article, you’d think this was about overbooking, but it’s not. (That’s just a buzzword that probably gets clicks.) This is really about revenue opportunities that don’t harm the customer experience. Whether this is a big enough opportunity for United to bother with is hard to say, but it’s worth talking about the concept.
The basic premise is simple. If United has a flight that’s sold out well before departure, this system would look for opportunities to offer vouchers to people to get them to switch to another flight. The process would start 5 days out, so this is well before anyone would get to the airport. If that happens, United could then re-sell those seats at a higher fare.
Though United gave me a short statement yesterday that would lead you to believe this about overbooking…
We are always looking at new ways to innovate and improve the customer experience and this extremely small test is an example of one of many opportunities we are reviewing. United has already taken steps to reduce overbooking, resulting in a nearly 90 percent year over year reduction of involuntary denied boardings for the month of June.
…I disagree. I actually agree with Dave Bartels, vice president pricing and revenue management for United, who said in the article that this isn’t really about overbooking.
If United were using this to reduce the level of overbooking, then it would effectively be admitting that its systems are incapable of accurately predicting by how much flights should be oversold. While President Scott Kirby has had plenty to say about United’s revenue management capabilities since he got there, he’s not going to rely on a patch like this to help fix the mess. He’s going to fix the root of the problem and make sure the revenue management system is doing its job well. The focus should be (and undoubtedly is) on finding the right algorithm to ensure that United is setting authorization levels (the number of tickets it sells on any given flight) appropriately.
Thanks to variability, it won’t be right on every flight, of course, and that’s why people are denied boarding. But if your system is running right, you don’t want to pay someone to change flights 5 days in advance of travel. You want to bet that through no-shows, missed connections, changes, and volunteers, your flights will go out with just the right number of people. (That’s why you decided to sell that many seats in the first place.) United should use a mechanism on the day of travel to get more volunteers if it finds itself in trouble then. It does that, and it’s developing more ways to handle that better… but that’s not what this tool is for. For this tool, let’s look at a simple example.
Let’s say you have 150 seats on a flight and your system determines that the right number of seats to sell considering no-shows, changes, etc. ends up being 160. Great. Now you historically expect to be able to sell 25 full fare tickets, 100 regular fare tickets, and 35 super-cheapo-fare tickets.
Here you are one week before departure, and demand was better than expected. In fact, you’ve sold all 160 seats, including the 25 full fare tickets that you were primarily holding back for last minute travelers. You sold too soon, and that’s something Scott Kirby noted was an issue at United. The airline is changing its processes so it doesn’t happen as often, but it will still happen. What can you do about that? Well, today you really have two options.
a) cry like a baby about the lost revenue and hope you can do better next time
b) increase authorization levels so you can sell more full fare tickets knowing that you’re highly likely to have to bump people
From a customer service (and decency) perspective, the right thing to do is go with option a). In that situation, this tool would allow United to go to someone 5 days in advance, offer them a little voucher to switch to a different flight, and then have enough time to re-sell that ticket to a last minute traveler paying full fare. Everyone is happy. United gets more revenue (assuming it actually can re-sell the seat). The volunteer gets to take a new flight that may even be better AND gets a voucher for it. Lastly, a last minute traveler who really wants to be on that airplane can buy a ticket. Huzzah.
But do we really think that airlines don’t use option b? In reality the system is constantly rebalancing, looking at the demand and adjusting inventory as it goes. In the past, it was probably far likelier that the system would open up more full fare seats if the flight was selling out knowing that United could get more money even accounting for the likelihood of having to pay someone to get bumped. But the climate has changed dramatically over the last few years, and airlines are getting much more pressure than before about NOT operating that way. But let’s say it still does happen…
In that case, this tool actually would reduce overbooking. And while everyone would still gain something, there are losses to be had as well. United doesn’t increase revenue since it was going to sell those tickets anyway, but it does decrease cost. Instead of having to pay hundreds in vouchers or cash to someone getting bumped on the day of travel it instead pays a whole lot less several days in advance. The volunteer may make less money on the deal, but he now earns that money way in advance and doesn’t have to get tortured during the airport process. For the last minute traveler, nothing changes and the seat is still available.
It’s hard to know how often United will find itself in a situation like this, especially once it gets it revenue management system working as it wants. If it happens very infrequently, then there’s nothing stopping United from doing this manually on its own today. But if it happens more often, then this could be a useful tool.
The theory is really quite compelling. If the technical pieces (primarily the analytical components) are genuinely combining current flight load and expected additional demand, load on adjacent flights, passenger-specific attributes (e.g. FF status) and origin and destination dynamics (e.g. frequency, business/leisure) this feels like a win-win proposition.
Honestly, as a loyal United flyer, I’d be very pleased to receive an offer of vouchers or miles or premier qualifying miles if I had the flexibility to move my flights.
I’m actually surprised they aren’t already doing something like this. Delta has offered me waivers on a number of occasions for changing an itinerary well in advance of the flight. Of course if there is no benefit to me I can keep everything as-is. Usually I have taken the offer to eliminate a connection or get a guaranteed upgrade to F on a different flight. If cash or vouchers were offered that would sweeten the deal but overall the concept doesn’t sound revolutionary to me.
It’s all about execution. United (as well as most carriers) have difficulty with some simple processes.
It really surprises me that airlines don’t already have some willingness to do this.
I had a personal experience with it a couple of years ago, with you guys, in fact. I got an email from Delta asking me about my willingness to bid for voluntary bumping from an overbooked flight, maybe one or two days in advance of the flight.
As it happened, my own plans had changed a bit. I called one of your concierges, and told him to call Delta and ask if they wanted to put me on a flight the night before (which was wide open) for no change fee. They declined. Never understood it.
This really does seem like it could benefit everybody.
grichard – It’s because the airlines historically don’t want to buy someone off if they aren’t sure that they’ll actually need the seat. That’s why they like doing it during check in. You get on a list, and once the count is in, they either pay you or board you. There’s no risk to them in having to actually pay out if they don’t need the seat.
I like the concept. In fact I’m on a business trip now where my plan is to fly home on Saturday, but my Friday appointments have cancelled. If I receive an email offering me to drop the Saturday flight for an earlier one home, I’d jump on it and save on the hotel and rental car expenses. In fact I’d even take mileage as compensation.
It makes a lot of sense, because there are times that demand just cannot be predicted and all of a sudden full fare tickets go flying off the shelf. Example: When the Cubs made the World Series, all of a sudden every seat from Chicago to Cleveland was magically transformed into a $250-$300 one way ticket. People paid it (anecdotally from my dad, his United flights were jammed) and I’m sure United would have loved to to do this to a few random students flying to Case Western Reserve who would not have had a problem flying during the game. Or you can do what I did and book a pile of speculative refundable Southwest tickets during the NLCS.
This is why I bought my ticket before the NLCS. And yes, those flights were 100% full
Robert – True, but in a very specific example like that, there’s nothing preventing United from manually sending an offer to people on those flights. (And they really should have!) It would have to be a broader issue for the software to make sense.
Further down the source article, it was noted that Quantas and Alaska are rolling out the same program later this summer.
Let’s apply the same concept to seat selection… ;-)
I just bought a ticket on Virgin America… when I went to the seat-selection map, I saw regular coach seats, and then seats farther forward marked “+$20.” I had no choice but to take one of those, because all the base-fare seats were taken. It was a pretty good fare anyway, but left a slight “got-nickel-and-dimed” feeling.
Insteat of only seeing plus signs (surcharges) on these seat maps, wouldn’t it be nice if the middle seats and bulkhead seats had, e.g. “-$15” marked on them? Give some nickels and dimes back to the pax!
You didn’t have to pay that $20. You could have simply not chosen a seat, and then at check-in, you would’ve been assigned a seat automatically at no cost.
Thanks for the tip… I hadn’t known that. So I’d have gotten a ‘premium’ seat w/o the surcharge (assuming no no-shows in the regular coach section)? Interesting outcome, and I suppose a one-in-three chance of a window!
There are typically some regular seats that are blocked until the day of, so you wouldn’t necessarily have gotten a free upgrade.
Ah, another piece of useful info… although personally, I care more about getting a window seat than a few extra inches of legroom. Being ‘only’ 5-10, I haven’t felt cramped in any carrier’s coach seat (except Frontier’s).
Anthony – Those premium seats on Virgin America aren’t actually premium. They’re the same exact seat as elsewhere. The only extra legroom seats they have are in Main Cabin Select and those are sold as a separate fare, not as a seat fee. So yeah, you would have just been put into a premium seat but you wouldn’t have known it.
Well, no wonder nobody bought them!
This may be the greatest Cranky graphic ever!
Wow Cranky, I’m impressed.
This makes total sense from a revenue management system. So long as the marginal revenue, in the form of a higher fare sold after the plane was sold out, exceeds the marginal expense of having to move el cheapo seat holders, this is an airline Nirvana.
You can selling up to time of flight with very high coach fares and then buy-off the el cheapos. That way you don’t miss out.
Instead of having to pay people to move off of full flights days in advance, the far better alternative and one every airline should be doing anyway is simply improve their revenue management forecast so that they accurately sell the correct number of seats at each fare level in advance. Domestic flights largely book within 2 months and the majority of domestic bookings outside of holidays is within one month before departure.
How do you know that they can “simply” do that?
I see many reasons why predicting no-shows/cancellation is an inexact science.
But if you can do it, pitch to them your algorithm as an alternative to this trial. :)
Airlines spend hundreds of millions of dollars on revenue management systems that are very accurate and are produced by a handful of companies. I believe the big 3 US airlines including United have built their own “black box” in the their revenue management systems because they don’t want any other airline to be able to copy their forecasting systems. United has talked about the lack of investment in UA’s revenue management technology so they know full well that they can do a better job of forecasting.
In the meantime, there are manual adjustments to any revenue management system that are possible to prevent taking too much discount traffic.
Further, airlines around the world have the ability to change their schedules including shifting aircraft between families (the 737-900/800 and A320/319 etc) in order to put seats where they can best be sold at the highest fare.
This whole program seems like an expensive and cumbersome process that doesn’t deal with the core problem which is that United isn’t accurately forecasting or managing its fleet at the 2-3 month period before departure so it needs to implement this just days before departure.
Tim – I agree completely on this. There is a core issue that United has acknowledge and this is just a band aid. But there is still always going to be variability, so there could be some instances where it would be a helpful tool. The question is whether there would be enough with a properly functioning revenue management system to make it worthwhile.
Interesting write up. NBC News report last night seemed to keep emphasizing “if it works”
Bulletin! New policy. “While we (UA) have a price list, you are all always welcome to ask for something less than list price. If we agree on a lesser than list-price, you get it. You save and we get the revenue we want. This policy applies before and after ticket purchase. Don’t worry about our revenue management system. We won’t worry about your income management system. Have a pleasant trip!”
Does United (still) wait list? Years ago, United did, and had a few priority levels. Perhaps they can use this to move a customer and accommodate a high priority wait-list customer. Great service to the priority customer and possibly higher yield for United.
You mean, standby waitlist? Sure.
No, confirmed reservation booking waitlist. At one time travel agencies and reservations agents could waitlist for a flight either by booking inventory ( typical situations were waitlist F, confirm Y or confirm Y, waitlist for lower fare bucket).
This test might provide an attractive opportunity if it didn’t require flying on United. Let’s hope the report that Alaska is going to implement the same type of program is accurate!
I noted that UA cited nearly 90% reduced YoY involuntary DBs in June.
I wonder how much of this was related to: a) increased volunteer-solicitation success and b) reduced total DBs. In turn, I wonder what drove changes in total DB’s: a) reduced demand, b) reduced overbooking, c) improved overbooking (aside from reductions), d) improved revenue integrity (also including reduced variability in show-up and late-cancellation rates).
In the most recent DOT reported data (1Q2017), UA had small improvements in total DB rate (from 8.9/10K to 8.2/10K) and in involuntary % (from 6.1% to 5.4%).
World Traveller – While we’ll probably never know, United did say it was reducing its overbooking levels as part of that 25,000 point plan after the Dr Dao thing. So that’s definitely a part of it. I would also assume that the big improvement in operational performance has made a difference too. But it also has to be process driven in handling the denied boardings as well. It takes a village…
In mathematical / finance terms, what you’re taking about is making an airline ticket ‘callable’ by an airline. The question about how to price this call option has been in discussion for some times – see the following 2 links from 2008 and 2010 talking about how to price such an option