Why doesn't AFKL offer booking guarantee on short/medium haul flights, why only in Y?
#16
Join Date: Sep 2000
Location: Berlin, Germany
Programs: DL DM/2MM, UA PE, HH Gold
Posts: 1,080
Sorry, but it's not. If the business cabin is full and an über-Ulti ( or worse, two of them! ) coming up exercising their booking guarantee someone has be to bumped. If you deny that somebody would have to be bumped then your "guarantee" is not worth much. Your guarantee means that i would get a seat on a flight that is already zeroed out in J, zero seats available. That makes it easier for airlines with a first class like LH by just op-up you to F, but what is KL supposed to do when YOU show up at the gate at an already full flight?
#17
Original Poster
Join Date: Apr 2005
Programs: Eurostar Carte Blanche, SBB-CFF-FFS GA-AG, SNCF Grand Voyageur LeClub
Posts: 7,836
Sorry, but it's not. If the business cabin is full and an über-Ulti ( or worse, two of them! ) coming up exercising their booking guarantee someone has be to bumped. If you deny that somebody would have to be bumped then your "guarantee" is not worth much. Your guarantee means that i would get a seat on a flight that is already zeroed out in J, zero seats available. That makes it easier for airlines with a first class like LH by just op-up you to F, but what is KL supposed to do when YOU show up at the gate at an already full flight?
You come back with the same point again: "but if the cabin is full..." - that's the point you miss. The airline doesn't fill the whole cabin before showing zero availability. There is inventory left, which can then be given to booking guarantee. Please read post #5 which explains how airlines do it (=not my invention, but this is how airlines manage their inventory every day). You will see that this is designed in a way not to bump anyone.
You only need to bump someone if there are more people turning up as forecasted, but that problem of the algo getting it wrong already exists today. I don't have the number of denied boardings for AFKL, but it's extremely small. So the accuracy of the forecasts is very good.
Also the question you ask about what KL should do if someone with booking guarantee shows up at the gate and the cabin is already filled with people. This has nothing to do with booking guarantee, which is, as explained, a mechanism that gives people a confirmed seat 24 or maybe 36 hours before departure. It's not about turning up at the gate and getting people off the plane.
Last edited by San Gottardo; Mar 29, 2019 at 7:17 am
#18
FlyerTalk Evangelist
Join Date: Mar 2008
Location: Netherlands
Programs: KL Platinum; A3 Gold
Posts: 28,733
They already overbook J. I think that it's rare to see any flight - or at least any route - unavailable to purchase in J more than 24 hours before departure.[I do recall someone posting here about the frequent lack of J seats on a particular AF flight from CAI to CDG several months in advance, though!]
As such, in 99.9999% of cases, you can buy a full-fare J ticket up to 24 hours before departure. And you can be sure you won't be the person bumped, if it comes to that.
As such, in 99.9999% of cases, you can buy a full-fare J ticket up to 24 hours before departure. And you can be sure you won't be the person bumped, if it comes to that.
#19
Join Date: Mar 2016
Location: CPT,AMS
Posts: 4,412
1) Overall? Very little I suppose, I guess it concentrates around some high travel periods. But I have been on flights where I was waitlisted and they told me they were 6 or 7 other people on the WL as well (this was not in IRROPS, where you can find yourself with a planeload-long wait list where people try to get on the next flight)
I think the most likely case for having 6 or 7 people in WL is that they themselves were involved in IRROPS, doesn't need to be major IRROPS, could be as simple as a missed connection due to slight delay.
2) Individually? Myself, three times this week with AF. Not that this happens every week, but just this week this resulted in 1 case where the wait list confirmed, 1 case where I flew on easyJet but 2 hours later than I needed with some unpleasant impact on my agenda, 1 case of just cancelling the entire trip and all the professional commitments that
#20
Original Poster
Join Date: Apr 2005
Programs: Eurostar Carte Blanche, SBB-CFF-FFS GA-AG, SNCF Grand Voyageur LeClub
Posts: 7,836
Err, well, that exactly is the point of my suggestion. I would welcome if the airline didn't subject me to that. I am happy to give them loads of money, but if for instance I do a day-trip to JFK and the outbound flight is full then I really don't want to do that in Y. No rest, not enough space to work, etc. Hence: please AFKL, do what other airlines do, offer booking guarantee also in J.
So what?
And again, I don't understand the question. This mechanism does exist and it works. So why are you trying to explain me, who merely points out the fact that it does exist and work elsewhere, that it doesn't work. Isn't that a bit futile? Reminds me of a discussion with someone in Germany some years ago, where I remarked that they didn't have credit-card check-out from parking lots (they always need to go to the machine and pay there before picking up the car). He also explained to me that this wouldn't work, it would mean that there are credit card readers at the exit, and what if someone forgot his card, etc. All kinds of reasons he came up with which either because he had never seen something like this or some remotely possible extreme situation. Completely ignoring the fact that credit card check out at parking exits existed and worked well all over France. Just doubting that something works only because it's unfamiliar, an interesting human trait.
Last edited by San Gottardo; Mar 29, 2019 at 9:10 am
#21
Join Date: Mar 2016
Location: CPT,AMS
Posts: 4,412
So then what do you suggest AF/KL do, put availability of 0.9?
And again, I don't understand the question. This mechanism does exist and it works. So why are you trying to explain me, who merely points out the fact that it does exist and work elsewhere, that it doesn't work. Isn't that a bit futile? Reminds me of a discussion with someone in Germany some years ago, where I remarked that they didn't have credit-card check-out from parking lots (they always need to go to the machine and pay there before picking up the car). He also explained to me that this wouldn't work, it would mean that there are credit card readers at the exit, and what if someone forgot his card, etc. All kinds of reasons he came up with which either because he had never seen something like this or some remotely possible extreme situation. Completely ignoring the fact that credit card check out at parking exits existed and worked well all over France. Just doubting that something works only because it's unfamiliar, an interesting human trait.
#22
Original Poster
Join Date: Apr 2005
Programs: Eurostar Carte Blanche, SBB-CFF-FFS GA-AG, SNCF Grand Voyageur LeClub
Posts: 7,836
I am not the person who writes the algos for AFKL, so I don't know what number the system spits out, just the probability, as you suggest, where a human then needs to make a decision and zero out availability; or where the system itself manages that inventory and takes a decision on when to zero out. I strongly suspect it is the later, and the bit I know about inventory management this is how it is done.
Now we have the conversation I was hoping we'd have , i.e. not whether this works (it does), but why AFKL doesn't do it. See thread title. I personally cannot think of any operational reason: their business model of a network carrier with long haul and short haul flights is identical to the LHLXOS, their fleet is also of a similar composition with some heavies, a large Airbus/737 fleet for Euro traffic, and some regional jets. It must be exactly that: they have in the past that customers don't value it. Hence my post, suggesting that actually passengers *do* value it. Not very many in absolute numbers, but those that are among the most loyal and most lucrative; loyalizing them further and doing so while being paid full fare is a win-win for the airline and for the passengers who are in situations where they are price-insensitive when they need to be in a certain place at a specific moment in time.
What works for one airline/country etc. doesn't always work for others (or not as efficient as), you are making an assumption here that AF/KL never even considered that option, I find that to be very unlikely and it is more likely they considered and rejected it, for whichever reasons.
#23
A FlyerTalk Posting Legend
Join Date: Sep 2009
Location: Minneapolis: DL DM charter 2.3MM
Programs: A3*Gold, SPG Plat, HyattDiamond, MarriottPP, LHW exAccess, ICI, Raffles Amb, NW PE MM, TWA Gold MM
Posts: 100,413
PMNW had a rule for a while (I don't remember the years) where Plats were guaranteed FULL Y seats 24 hours in advance and FULL J 7 days in advance IIRC, on any worldwide NW operated flight. I once asked how this would work if the flight were overbooked and whether as the last ticket purchased I would be on the top of the list to be bumped. The response was that NW would pick someone else to IDB if needed, basically paying someone else to not take the flight or be downgraded so that I would get the seat I had booked under the program.
At some point very bad weather was going to cause IROPs the next day (causing an overnight delay in my case) on a TATL WBC ticket and I called the NW Plat line (Chisholm) to try to re-route myself even though my flights had not yet been posted with delays. After conferring with a supervisor, the Plat agent agreed to rebook me onto a different TATL flight and told me to just get to the airport early to have the ticket changed, etc., as everything had been documented. I did so, and then boarded when WBC was called. Along the way a GA said that my seatmate and I were similar cases, but with him being Gold while I was Plat, she had given me the aisle seat and him the window. During PDBs we were chatting a bit, but then stopped when a couple who were boarding loudly pointed to us and talked to each other about how we had taken their seats and they had been kicked back to coach. We just looked at each other and smiled, although it seemed from the conversation we couldn't help overhearing that they had been well compensated for the downgrade and weren't really unhappy about the situation.
At some point very bad weather was going to cause IROPs the next day (causing an overnight delay in my case) on a TATL WBC ticket and I called the NW Plat line (Chisholm) to try to re-route myself even though my flights had not yet been posted with delays. After conferring with a supervisor, the Plat agent agreed to rebook me onto a different TATL flight and told me to just get to the airport early to have the ticket changed, etc., as everything had been documented. I did so, and then boarded when WBC was called. Along the way a GA said that my seatmate and I were similar cases, but with him being Gold while I was Plat, she had given me the aisle seat and him the window. During PDBs we were chatting a bit, but then stopped when a couple who were boarding loudly pointed to us and talked to each other about how we had taken their seats and they had been kicked back to coach. We just looked at each other and smiled, although it seemed from the conversation we couldn't help overhearing that they had been well compensated for the downgrade and weren't really unhappy about the situation.
#24
Join Date: Jul 2015
Location: HAG
Programs: Der 5* FTL
Posts: 8,052
Now we have the conversation I was hoping we'd have , i.e. not whether this works (it does), but why AFKL doesn't do it. See thread title. I personally cannot think of any operational reason: their business model of a network carrier with long haul and short haul flights is identical to the LHLXOS, their fleet is also of a similar composition with some heavies, a large Airbus/737 fleet for Euro traffic, and some regional jets. It must be exactly that: they have in the past that customers don't value it. Hence my post, suggesting that actually passengers *do* value it. Not very many in absolute numbers, but those that are among the most loyal and most lucrative; loyalizing them further and doing so while being paid full fare is a win-win for the airline and for the passengers who are in situations where they are price-insensitive when they need to be in a certain place at a specific moment in time.
The multitude of risks of overbooking J unintentionally has been raised, as has impact of keeping seats intentionally free "just in case". Also the fact that the routes where this guarantee would be most likely to be used are the routes where AFKL can least afford to either keep extra seats empty "just in case" or pissing off their existing customers. I can't see why AFKL would not have considered this and decided the benefits are not worth it.
As per comparison with LHG, the scale is a bit different in the end. LHG operates through 4 "main" hubs, of which 3 have at least some F service, of which 1 has F service on all longhauls, plus assorted other hubs and focus cities like GVA, BRU and wherever EW is parking its longhaul fleet today. Meanwhile AFKL has 2 "main" hubs, of which only 1 has some F service and none has full F service, plus 1 other minihub, which is also a co-terminal.
The potential for solving the overbooking situations by upgrading or rerouting is barely comparable. The potential impact of IRROPS or predicted IRROPS (potential for a larger amount of high elites to try and use the guarantee to get out/in before the meltdown) is also considerably higher.
I consider the comparison only very notionally valid.
#25
Join Date: Jun 2005
Location: 🇸🇬 🇭🇰 🇫🇷
Programs: Many
Posts: 4,749
It's interesting to see how people treat this as if it was a new invention and come up with all kinds of reasons why this wouldn't work - when in fact it does exist, it works, it's just that AFKL hasn't introduced it yet. The question is not "would this work" (we know it does), the question is "If it works elsewhere, why shouldn't AFKL also offer it?"
I think we can discuss ad nauseam about how good algos are (they are) whether the guaranteed Y class seat benefit model is transposable to J class (it is).
To me the rational answer to that question is rather... emotional and cultural: AF is not pragmatic and (premium) customer oriented when it comes to their international J. So any innovation that could trigger a risk is moved aside.
#26
Original Poster
Join Date: Apr 2005
Programs: Eurostar Carte Blanche, SBB-CFF-FFS GA-AG, SNCF Grand Voyageur LeClub
Posts: 7,836
For those that don't believe that it exists or understand how it works: this video
Booking guarantee is exactly that: letting the inventory management system set availability for certain buckets, by simply adding one bucket which is used for people that book last minute.
I still fail to see why people think that introducing booking guarantee suddenly creates a problem of bumping pax. Inventory management is an existing and proven approach, using existing and proven instruments, with existing data.
Also the fact that the routes where this guarantee would be most likely to be used are the routes where AFKL can least afford to either keep extra seats empty "just in case" or
And why do do you think that seats would go out empty?
pissing off their existing customers.
I can't see why AFKL would not have considered this and decided the benefits are not worth it.
As per comparison with LHG, the scale is a bit different in the end. LHG operates through 4 "main" hubs, of which 3 have at least some F service, of which 1 has F service on all longhauls, plus assorted other hubs and focus cities like GVA, BRU and wherever EW is parking its longhaul fleet today. Meanwhile AFKL has 2 "main" hubs, of which only 1 has some F service and none has full F service, plus 1 other minihub, which is also a co-terminal.
- And you believe that the inventory algos that AFKL uses today are not taking that into account? We can be certain that they do inventory management with the inventory that they have
- Not sure how more F seats make a difference for the short haul network
The potential for solving the overbooking situations by upgrading or rerouting is barely comparable. The potential impact of IRROPS or predicted IRROPS (potential for a larger amount of high elites to try and use the guarantee to get out/in before the meltdown) is also considerably higher.
Still puzzled how people accept the fact that airlines do yield and inventory management, accept the fact that airlines overbook flights so as to optimise yield - but when the suggestion comes along to merely create another bucket in the inventory those same people question the very workings of inventory management.
Very valid question. We must assume it indeed works since LH Group has implemented it and is keeping this benefit. Still they remain an exception in the game.
I think we can discuss ad nauseam about how good algos are (they are) whether the guaranteed Y class seat benefit model is transposable to J class (it is).
To me the rational answer to that question is rather... emotional and cultural: AF is not pragmatic and (premium) customer oriented when it comes to their international J. So any innovation that could trigger a risk is moved aside.
I think we can discuss ad nauseam about how good algos are (they are) whether the guaranteed Y class seat benefit model is transposable to J class (it is).
To me the rational answer to that question is rather... emotional and cultural: AF is not pragmatic and (premium) customer oriented when it comes to their international J. So any innovation that could trigger a risk is moved aside.
#27
Join Date: Jul 2015
Location: HAG
Programs: Der 5* FTL
Posts: 8,052
What makes you think that? I do not know it. Which is why this thread title is a question: "Why doesn't AFKL offer booking guarantee on short/medium haul flights, why only in Y?"
Yes. And - sorry to say it that bluntly - they are wrong. It basically refuses to recognize that airlines do inventory and revenue management, it basically refuses to recognize that airlines use different fare buckets to make them available to different markets/customer segments, it refuses tomrecognise the fact that airlines overbook their flights using forecasting methods in such a way that they sell more seats than there are on a plane but still almost never have to bump someone.
For those that don't believe that it exists or understand how it works: this video https://youtu.be/i7OgtWAdlsU explains how airlines do forecasting, protect certain fare classes for certain segments of customers, use nested inventory.
For those that don't believe that it exists or understand how it works: this video https://youtu.be/i7OgtWAdlsU explains how airlines do forecasting, protect certain fare classes for certain segments of customers, use nested inventory.
Booking guarantee is exactly that: letting the inventory management system set availability for certain buckets, by simply adding one bucket which is used for people that book last minute.
I still fail to see why people think that introducing booking guarantee suddenly creates a problem of bumping pax. Inventory management is an existing and proven approach, using existing and proven instruments, with existing data.
I still fail to see why people think that introducing booking guarantee suddenly creates a problem of bumping pax. Inventory management is an existing and proven approach, using existing and proven instruments, with existing data.
You seem to think current inventory management is good enough for that, I think it's not.
At the same point, that is also the flight where AFKL faces the highest lost revenue if the last seat is not sold.
Your solution to the capacity issue was to keep some seats in a separate bucket and release them only if unsold at point of last guarantee. That could work, but only if you assume 100% reliable inventory control and/or some reliable very, very last minute demand (<24h).
Wouldn't it be safe to assume that we simply don't know whether they haven't looked at it (didn't have the idea that this was something worthwhile looking into) or did look at it and came to the conclusion that their pax wouldn't value it? Maybe you do know, I at least don't.
I rather presume it's more likely they did.
I am not talking about short haul network. The whole discussion is about the longhaul business guarantee
Anwyay the point in relation to LHG here is, that regulating IRROPS by rerouting via 3/4 hubs is a lot, lot easier than rerouting via 1/2 hubs, lessening the commercial impact.
Still puzzled how people accept the fact that airlines do yield and inventory management, accept the fact that airlines overbook flights so as to optimise yield - but when the suggestion comes along to merely create another bucket in the inventory those same people question the very workings of inventory management.
Exactly my point. We know that it isn't an inventory management problem (the likelihood of people being bumped does not go up compared to today). Unless AFKL's inventory management is inferior to LHG's - which nobody has a reason to believe. But I can only speculate why AFKL doesn't offer it. My hypothesis is the same as yours, it's cultural and a different business philosophy. Fair enough
Cultural factors, that I don't want to say there is no impact of them on the decision. after all the simple look at tiers with both benefits and ease/difficulty of achieving suggests, that LH values their high-tier flyers relatively more compared to paid J pax in general in comparison to AFKL.
#28
Original Poster
Join Date: Apr 2005
Programs: Eurostar Carte Blanche, SBB-CFF-FFS GA-AG, SNCF Grand Voyageur LeClub
Posts: 7,836
We largely agree, have a few misunderstandings - and there is one point where you lost me.
Huh? Not at all.
There are *two* things which I believe you have mixed up in one statement:
1) What indeed I am certain to be right: "They could do it". aArlines with sophisticated inventory management have the possibility to manage inventory for certain customer segments by using fare buckets. They do that successfully today. The algorithms doing that are not 100,00% fail-proof, but close enough for airlines to feel comfortable with the very small number of IBD that creates. That is an undisputable fact, and I keep repeating it when it is being put in question. On this point, only on this, yes, I maintain, claiming the opposite is wrong. I would love to move past that point so that the second one, on which I have no fixed option at all but rather ask a question, can be discussed:
2) I *ask the question* why AFKL doesn't do it. There has been very little discussion about that, although that is the actual question that I put out. The few statements I made were that neither do I know whether they have considered it or not. So I am not saying that anybody is right or wrong or that AFKL is wrong not offering it - I just wonder why they don't. And I point out that some people would find it valuable.
Even if I repeat myself: I repeat that I do not know whether AFKL knows about the existence or doesn't. I do accept the possibility that they know it exists.
I concur. Maybe they have come to that conclusion after consideration. Or maybe they haven't considered it (you said in your last post that you were sure that they had considered it. You seem to be sure, I am not. Not a problem)
Well good, if you are so familiar with it, why are we having the debate about booking guarantee creating more IDBs.
Precisely, that is how inventory management works! That is already how they manage their fare buckets today between J, C, D, Z, I and so on: they close lower booking classes so that they can keep higher ones open. At one point they close them all to avoid creating too much overbooking and having to bump people. Nothing to do with booking guarantee. For instance, if the algo gets it wrong and closes I, Z, D, and C too early and only leaves seats in J because the airline expects people to turn up and buy J but then they don't - unsold seats are the result. And vice versa, if you leave availability open for too long (=create too much overbooking) then you create IDBs. We agree on that, don't we? Now just apply exactly that principle to booking guarantee: make the booking guarantee class the highest booking class (with the highest protection level), then close lower booking classes earlier depending on what you expect for the highest booking class. Exactly the same thing as today. Booking guarantee is just a different bucket, the one that you close last.
?? This is the part I don't get. You accept that inventory management is good enough to manage inventory between the different fare buckets, that it manages to close fare buckets earlier or later based on predictions, differentiating between different protection levels - but it wouldn't be good enough to manage one additional bucket using precisely the same methods? Why not?
I agree. In today's world, this is why they leave lower booking classes open long enough toavoid unsold seats. In the booking guarantee world, they would leave "non-Elite" availability open long enough to avoid unsold seats.
To be precise: "leave unsold or steer overbooking in such a way as to be able to accommodate these pax". But yes, that is how it works. Just like the highest fare classes today have the highest protection levels and are the ones closed last.
You are right. You would need 100% reliable inventory management to avoid unsold seats or to avoid overbooking. But that is the case already today. See above: if you close lower booking classes too early you may find yourself with unsold seats if you mis-predict the number of people buying at high fares; or if you leave booking classes open for too long , you create too much overbooking. But that is already a problem with inventory management today, predictions can be wrong. Not different if you introduce booking guarantee. You can get it wrong as well in the future, but there is no reason why you'd get it wronger than before.
In theory, yes. But, the way this was explained to me by a revenue manager is that you hardly have any more fluctuation in the last 24 hours before departure. The ratio of booked seats/available seats (the "overbooking factor") stays pretty stable, and waitlists hardly move.
This is where I fundamentally disagree with you. I believe that over time repeat high fare customers have higher economic value for the company than one-time customers. I would look at LTCV (long term customer value or life time customer value, depending on whose term you use) rather than the economic value a customer generates with one flight. Especially since the booking guarantee customer would in any case have to pay the highest fare. Result: the booking guarantee customer has higher lifetime value and equal transaction-value-on-this-flight.
Whether Plats are always higher LTCV customers is a different debate and may well be questioned.
No. What I have certainty about is that inventory management systems can run a booking guarantee scheme without problems. That is one statement.
What I have no certainty about at all and said so repeatedly and therefore asked the question was if they had considered and dismissed it or never considered it. I don't know. That is another statement.
Yes. Because I can think of some customers who find this valuable.
I accept that. I have absolutely no basis for believing the same or refuting it. Maybe you have some evidence. I just don't know.
Because you mention another airline's inventory.
Is it? Since when? Not in my original quesiton at least. Because it's also in the short haul network that I would find booking guarantee super helpful and where it is sometimes needed.
Yes, you are right. This is exactly the situation where bumping pax will be unavoidable and it will cost the airline. The very same moment where those that can benefit from the booking guarantee will say "Those guys really got me out of a critical situation". In marketing there is something called the gratitude moment which translates into higher loyalty, which is worth money (FYI: there are firms who actively play with that. Make small issues appear big and then let the customer agent who solves the problem appear as the saviour, and create the moment of gratitude, which has been proven to translate into higher lifetime value of those customers)
But there are ways around it: AFKL could suspend booking guarantee in the case of network meltdowns (strikes in France...) and simply revert to the promise that already exists today: "you'll be top of the wait list."
Why? Hypothetical example: ff Lufthansa has to bump a passenger from Dublin to Athens; it can put him on one of four LHG A320 planes that leave in the next couple of hours. Each of those next planes has 160 seats, a total of 640 seats. One connections is via Frankfurt, one via Munich, one via Zurich, one via Vienna. To re-accommodate the bumped pax it has to find just one seat among the 640 seats that leave in the next hours.
Now if KLM has to bump a passenger from Dublin to Athens; it can put him on one of four A320/B738 AFKL planes that leave in the next couple of hours. Each of those next planes has 160 seats, a total of 640 seats. Three connections are via Amsterdam, one via Paris. To re-accommodate the bumped pax it has to find just one seat among the 640 seats that leave in the next hours.
The point that I am trying to make: it doesn't matter if you have four hubs or two or three. What matters is how much capacity you have to re-accommodate bumped passengers. It doesn't matter if all those re-accommodation solutions fly through one or different hubs.
The debate would then be if LHG has a much denser network than AFKL which would allow it to have access to more inventory. Probably the answer depends on what connections one is looking at, and to what extent airline alliances are being taken into account. Being bumped of a Lufthansa flight from ATL or Rio de Janeiro will leave you in a worse place than bumped off an AF flight. LHG has one daily to each ATL and GIG, whilst AFKL have three dailies to ATL (plus the DL JV) and 18 weekly flights to GIG. Bangkok would be the other way around: one AF and one KL, but three (or even four?) daily LHG flights and alliance partner TG.
They do. Already today. Also in Business Class. Nothing news. Algos are imperfect.
You would be right if booking guarantee worked until the last minute, because as you say it would remove the concept of sold out. But that's not what I suggested. The flight *is* sold out even for everybody, including the top customer 24/36 hours (pick your number) before departure. I just apply the same principle that inventory already apply today: close lower booking classes earlier to only leave the higher ones open. I just make the booking guarantee class the highest one.
Yes, we agree.
Originally Posted by [url=http://fabo.sk/
Fabo.sk;30960520]Because, respectfully, you keep repeating "someone else does it, I think they should do it, therefore they are wrong and I am right"
There are *two* things which I believe you have mixed up in one statement:
1) What indeed I am certain to be right: "They could do it". aArlines with sophisticated inventory management have the possibility to manage inventory for certain customer segments by using fare buckets. They do that successfully today. The algorithms doing that are not 100,00% fail-proof, but close enough for airlines to feel comfortable with the very small number of IBD that creates. That is an undisputable fact, and I keep repeating it when it is being put in question. On this point, only on this, yes, I maintain, claiming the opposite is wrong. I would love to move past that point so that the second one, on which I have no fixed option at all but rather ask a question, can be discussed:
2) I *ask the question* why AFKL doesn't do it. There has been very little discussion about that, although that is the actual question that I put out. The few statements I made were that neither do I know whether they have considered it or not. So I am not saying that anybody is right or wrong or that AFKL is wrong not offering it - I just wonder why they don't. And I point out that some people would find it valuable.
without accepting the possibility, however remote, that AFKL actually does know this is a thing that exists,
and maybe, just maybe, they have decided that it's not in the best interest of their business to offer this.
With all due respect, I know that inventory management exists. I wasn't born yesterday.
Because the only way you can prevent bumping pax is knowing ahead of time exactly what flight the booking guarantee is going to be used and not sell the exact number of seats that will be filled by booking guarantee using pax... or else keep entirely too much reserve inventory free, which will prevent you from selling the available seats at close-to-full fare anyway.
You seem to think current inventory management is good enough for that, I think it's not.
The only point when a booking guarantee is needed is on a flight that would otherwise be full.
At the same point, that is also the flight where AFKL faces the highest lost revenue if the last seat is not sold.
At the same point, that is also the flight where AFKL faces the highest lost revenue if the last seat is not sold.
Your solution to the capacity issue was to keep some seats in a separate bucket and release them only if unsold at point of last guarantee.
That could work, but only if you assume 100% reliable inventory control
and/or some reliable very, very last minute demand (<24h).
You're expecting that AFKL considers business class pax on high-demand routes low-value customers. Just because a pax doesn't have AFKL Platinum card, doesn't mean they aren't valuable.
Whether Plats are always higher LTCV customers is a different debate and may well be questioned.
You say you don't, but you act like you know they didn't
What I have no certainty about at all and said so repeatedly and therefore asked the question was if they had considered and dismissed it or never considered it. I don't know. That is another statement.
and you want them to.
I rather presume it's more likely they did.
Yes, certainly the inventory and revenue management takes these factors into account. How is that even relevant?
I am not talking about short haul network. The whole discussion is about the longhaul business guarantee
Which only makes the problem worse. While it's possible to predict loads fairly well in regular operations, it is impossible to predict when meltdowns will happen exactly. So you might aim to have 1-2 empty seats at T-48 and get to that goal, but suddenly the French ATC announce a strike and you have 10 Plats trying to make use of the booking guarantee, which leaves you with 8 people who already bought the highest cabin available that you need to bump or reroute. Rerouting someone on dirt-cheap eco ticket has far lower impact than rerouting someone in business class.
But there are ways around it: AFKL could suspend booking guarantee in the case of network meltdowns (strikes in France...) and simply revert to the promise that already exists today: "you'll be top of the wait list."
Anwyay the point in relation to LHG here is, that regulating IRROPS by rerouting via 3/4 hubs is a lot, lot easier than rerouting via 1/2 hubs, lessening the commercial impact.
Now if KLM has to bump a passenger from Dublin to Athens; it can put him on one of four A320/B738 AFKL planes that leave in the next couple of hours. Each of those next planes has 160 seats, a total of 640 seats. Three connections are via Amsterdam, one via Paris. To re-accommodate the bumped pax it has to find just one seat among the 640 seats that leave in the next hours.
The point that I am trying to make: it doesn't matter if you have four hubs or two or three. What matters is how much capacity you have to re-accommodate bumped passengers. It doesn't matter if all those re-accommodation solutions fly through one or different hubs.
The debate would then be if LHG has a much denser network than AFKL which would allow it to have access to more inventory. Probably the answer depends on what connections one is looking at, and to what extent airline alliances are being taken into account. Being bumped of a Lufthansa flight from ATL or Rio de Janeiro will leave you in a worse place than bumped off an AF flight. LHG has one daily to each ATL and GIG, whilst AFKL have three dailies to ATL (plus the DL JV) and 18 weekly flights to GIG. Bangkok would be the other way around: one AF and one KL, but three (or even four?) daily LHG flights and alliance partner TG.
Yes, and that's why IDBs never happen... unless they do.
Maybe I don't give AFKL inventory management enough credit, but I don't think the likelyhood of people being bumped stays the same. After all, you are removing the concept of "sold out" for a certain class of customers.
Cultural factors, that I don't want to say there is no impact of them on the decision. after all the simple look at tiers with both benefits and ease/difficulty of achieving suggests, that LH values their high-tier flyers relatively more compared to paid J pax in general in comparison to AFKL.
#29
Join Date: Jul 2015
Location: HAG
Programs: Der 5* FTL
Posts: 8,052
This is a good discussion.
So far so good. I did get a feeling that you think they didn't have considered it and if they did, they would have offered it. But of course, I can be wrong.
Because it is necessary that a booking guarantee mechanism increases potential IDBs, as I shall follow on:
The key difference is, that even if you leave availability open for too long, you might easily end up with lower income due to lower fares getting sold, but you can still zero out all classes at capacity (or capacity +1 or capacity +2, as you desire, whether that is specific flight capacity, capacity for the rest of the day, group capacity etc.). General guarantee mechanism (i.e. open to all eligible customers at all times) removes this option as it forces theoretically up to unrestricted (but always non-zero) availability.
Eventually, someone will use this non-zero availability to force an (undesirable) overbooking (which is not much of a problem in Y, but is a much bigger issue in J).
Because I believe that booking guarantee demand is much more volatile than general demand and as such less predictable.
I would say the reason is necessarily increased volatility of guaranteed flight demand + inability to zero out a flight in advance.
I would also not expect this to actually exist in the wild, but it would be an option, if it existed.
That is an entirely fair belief. Probably less likely to be shared by AFKL, judging by signals such as the state of Ultimate program...
I would still say it's riskier than not having such a scheme; that however does not mean it can not be worth it to an airline. I just think AFKL has decided, opposite to LHg (and in line with most other airlines around the world, it seems) that it isn't.
Some? Certainly. Enough? That'd debatable.
As a point of comparison of what can drive the difference in risk assesment. Certainly general revenue management takes the structure in account in both cases.
OK, maybe I shouldn't have used the term "whole discussion", although to be fair, only one other poster also mentioned short haul, in reaction to your reaction to them, and to point out that they don't disagree with you.
That is actually an extremely good point I haven't considered...
Yes, potentially, or limit the number of "guarantee" seats on any given flight to, say, 2... however, that sort of defeats the purpose of the guarantee I'm afraid, making it sort of a "we'll take you if we have place" stand-in with only slightly better conditions for qualifying elites than general public.
The difference is, if there is flight disruption in Amsterdam, you are down to 1/4 capacity. If there is a disruption in Frankfurt, LH still has 3/4 capacity. Or maybe 1/2 of F seats, if we take a hypothetical destination served from FRA and ZRH with F and MUC and VIE with no F.
Not particularly more inventory, but more distributed, and therefore more disruption resistant.
Yes, but a guarantee concept would make the imperfection even more pronounced, IMO.
It removes the concept of "sold out" at any time sooner than 24h before departure. And as you earlier yourself noted, there is not much sales in this period anyway...
We largely agree, have a few misunderstandings - and there is one point where you lost me.
Huh? Not at all.
There are *two* things which I believe you have mixed up in one statement:
1) What indeed I am certain to be right: "They could do it". aArlines with sophisticated inventory management have the possibility to manage inventory for certain customer segments by using fare buckets. They do that successfully today. The algorithms doing that are not 100,00% fail-proof, but close enough for airlines to feel comfortable with the very small number of IBD that creates. That is an undisputable fact, and I keep repeating it when it is being put in question. On this point, only on this, yes, I maintain, claiming the opposite is wrong. I would love to move past that point so that the second one, on which I have no fixed option at all but rather ask a question, can be discussed:
2) I *ask the question* why AFKL doesn't do it. There has been very little discussion about that, although that is the actual question that I put out. The few statements I made were that neither do I know whether they have considered it or not. So I am not saying that anybody is right or wrong or that AFKL is wrong not offering it - I just wonder why they don't. And I point out that some people would find it valuable.
Huh? Not at all.
There are *two* things which I believe you have mixed up in one statement:
1) What indeed I am certain to be right: "They could do it". aArlines with sophisticated inventory management have the possibility to manage inventory for certain customer segments by using fare buckets. They do that successfully today. The algorithms doing that are not 100,00% fail-proof, but close enough for airlines to feel comfortable with the very small number of IBD that creates. That is an undisputable fact, and I keep repeating it when it is being put in question. On this point, only on this, yes, I maintain, claiming the opposite is wrong. I would love to move past that point so that the second one, on which I have no fixed option at all but rather ask a question, can be discussed:
2) I *ask the question* why AFKL doesn't do it. There has been very little discussion about that, although that is the actual question that I put out. The few statements I made were that neither do I know whether they have considered it or not. So I am not saying that anybody is right or wrong or that AFKL is wrong not offering it - I just wonder why they don't. And I point out that some people would find it valuable.
So far so good. I did get a feeling that you think they didn't have considered it and if they did, they would have offered it. But of course, I can be wrong.
Precisely, that is how inventory management works! That is already how they manage their fare buckets today between J, C, D, Z, I and so on: they close lower booking classes so that they can keep higher ones open. At one point they close them all to avoid creating too much overbooking and having to bump people. Nothing to do with booking guarantee. For instance, if the algo gets it wrong and closes I, Z, D, and C too early and only leaves seats in J because the airline expects people to turn up and buy J but then they don't - unsold seats are the result. And vice versa, if you leave availability open for too long (=create too much overbooking) then you create IDBs. We agree on that, don't we? Now just apply exactly that principle to booking guarantee: make the booking guarantee class the highest booking class (with the highest protection level), then close lower booking classes earlier depending on what you expect for the highest booking class. Exactly the same thing as today. Booking guarantee is just a different bucket, the one that you close last.
Eventually, someone will use this non-zero availability to force an (undesirable) overbooking (which is not much of a problem in Y, but is a much bigger issue in J).
?? This is the part I don't get. You accept that inventory management is good enough to manage inventory between the different fare buckets, that it manages to close fare buckets earlier or later based on predictions, differentiating between different protection levels - but it wouldn't be good enough to manage one additional bucket using precisely the same methods? Why not?
You are right. You would need 100% reliable inventory management to avoid unsold seats or to avoid overbooking. But that is the case already today. See above: if you close lower booking classes too early you may find yourself with unsold seats if you mis-predict the number of people buying at high fares; or if you leave booking classes open for too long , you create too much overbooking. But that is already a problem with inventory management today, predictions can be wrong. Not different if you introduce booking guarantee. You can get it wrong as well in the future, but there is no reason why you'd get it wronger than before.
In theory, yes. But, the way this was explained to me by a revenue manager is that you hardly have any more fluctuation in the last 24 hours before departure. The ratio of booked seats/available seats (the "overbooking factor") stays pretty stable, and waitlists hardly move.
This is where I fundamentally disagree with you. I believe that over time repeat high fare customers have higher economic value for the company than one-time customers. I would look at LTCV (long term customer value or life time customer value, depending on whose term you use) rather than the economic value a customer generates with one flight. Especially since the booking guarantee customer would in any case have to pay the highest fare. Result: the booking guarantee customer has higher lifetime value and equal transaction-value-on-this-flight.
Whether Plats are always higher LTCV customers is a different debate and may well be questioned.
Whether Plats are always higher LTCV customers is a different debate and may well be questioned.
No. What I have certainty about is that inventory management systems can run a booking guarantee scheme without problems. That is one statement.
What I have no certainty about at all and said so repeatedly and therefore asked the question was if they had considered and dismissed it or never considered it. I don't know. That is another statement.
What I have no certainty about at all and said so repeatedly and therefore asked the question was if they had considered and dismissed it or never considered it. I don't know. That is another statement.
As a point of comparison of what can drive the difference in risk assesment. Certainly general revenue management takes the structure in account in both cases.
Yes, you are right. This is exactly the situation where bumping pax will be unavoidable and it will cost the airline. The very same moment where those that can benefit from the booking guarantee will say "Those guys really got me out of a critical situation". In marketing there is something called the gratitude moment which translates into higher loyalty, which is worth money (FYI: there are firms who actively play with that. Make small issues appear big and then let the customer agent who solves the problem appear as the saviour, and create the moment of gratitude, which has been proven to translate into higher lifetime value of those customers)
Why? Hypothetical example: ff Lufthansa has to bump a passenger from Dublin to Athens; it can put him on one of four LHG A320 planes that leave in the next couple of hours. Each of those next planes has 160 seats, a total of 640 seats. One connections is via Frankfurt, one via Munich, one via Zurich, one via Vienna. To re-accommodate the bumped pax it has to find just one seat among the 640 seats that leave in the next hours.
Now if KLM has to bump a passenger from Dublin to Athens; it can put him on one of four A320/B738 AFKL planes that leave in the next couple of hours. Each of those next planes has 160 seats, a total of 640 seats. Three connections are via Amsterdam, one via Paris. To re-accommodate the bumped pax it has to find just one seat among the 640 seats that leave in the next hours.
The point that I am trying to make: it doesn't matter if you have four hubs or two or three. What matters is how much capacity you have to re-accommodate bumped passengers. It doesn't matter if all those re-accommodation solutions fly through one or different hubs.
Now if KLM has to bump a passenger from Dublin to Athens; it can put him on one of four A320/B738 AFKL planes that leave in the next couple of hours. Each of those next planes has 160 seats, a total of 640 seats. Three connections are via Amsterdam, one via Paris. To re-accommodate the bumped pax it has to find just one seat among the 640 seats that leave in the next hours.
The point that I am trying to make: it doesn't matter if you have four hubs or two or three. What matters is how much capacity you have to re-accommodate bumped passengers. It doesn't matter if all those re-accommodation solutions fly through one or different hubs.
The debate would then be if LHG has a much denser network than AFKL which would allow it to have access to more inventory. Probably the answer depends on what connections one is looking at, and to what extent airline alliances are being taken into account. Being bumped of a Lufthansa flight from ATL or Rio de Janeiro will leave you in a worse place than bumped off an AF flight. LHG has one daily to each ATL and GIG, whilst AFKL have three dailies to ATL (plus the DL JV) and 18 weekly flights to GIG. Bangkok would be the other way around: one AF and one KL, but three (or even four?) daily LHG flights and alliance partner TG.
You would be right if booking guarantee worked until the last minute, because as you say it would remove the concept of sold out. But that's not what I suggested. The flight *is* sold out even for everybody, including the top customer 24/36 hours (pick your number) before departure. I just apply the same principle that inventory already apply today: close lower booking classes earlier to only leave the higher ones open. I just make the booking guarantee class the highest one.
#30
Original Poster
Join Date: Apr 2005
Programs: Eurostar Carte Blanche, SBB-CFF-FFS GA-AG, SNCF Grand Voyageur LeClub
Posts: 7,836
Yes, thank you for participating. I won't be quoting each statement anymore, just to make it more legible, as we seem to agree on the majority of points anyway.
Forgive me for being daft, but I don't understand what you are saying. Not intellectually, just language-wise. Not saying if it's right or wrong, I just don't understand the sentence. But it sounds like the key argument, could you possible re-phrase it, as I'd really like to understand before expressing an opinion. Thx.
Why would booking guarantee demand be more volatile? Is there something in the demand pattern that makes this so? I don't believe so.
As we both agree, accuracy of predictions is key to maximise the yield, but the airline already has that information which they use today to make predictions and steer the availability of different fare buckets. What would change: operationalise it in such a way as to leave a nested inventory for those last minute bookers.
Yes. That is the core of the discusison I was trying to kick off. And suggest to AF that they could have a little more of such thinking.
Let's for the sake of discussion there weren't that many. That would then go in the sense of your argument: if only very few people made use of the booking guarantee then there necessarily would only be few situations where they'd upset the booking situation.
But then, what is "enough"?
Well... it is in the thread title, the OP and I clarified it again later. I'll show what I mean in the next post.
Yes, you are right for the IRROPS scenarios where there is one hub impacted (e.g. strike in France) and you have a hord of Elites use their booking guarantee to book themselves on the flights that leave through the other hub. But applying booking guarantee during normal operations i) there usually isn't any bumping anyway ii) if it happens then it really comes down to finding a seat on any flight, irrespective of the connecting hub.The fact that there are extraordinary situations such as the one you describe should not be a reason not to introduce the system. That would be throwing out the baby with the bathwater. Instead, suspend booking guarantee. For instance, when AF is hit by a strike and they communicate which flights will operate and which note they can just add a note saying "booking guarantee will be suspended during the strike, but Elite customers will retain the benefit of being top of the waiting list".
The point goes the other way around: you have to stop booking guarantee about 24/36 hours before departure because the people that by H-24 are still booked on the flight are much more likely to turn up than all the people at say H-72. If you look how overbooking factors develop over time: they are much higher further away from departure, and reduce so as to converge almost with actual seat capacity.
In the end it comes back to the same principle: airline closes lower booking classes at a point in time to always have enough seats left in the highest booking class for the number of people that it predicts to take that highest booking class (BTW the video link I posted explains that really well). All I am suggesting is to make that highest booking class one that is only accessible to elites, only when all other classes are zeroed out, and only up to 24/36 hours before departure. Is it thinkable to have certain booking classes or seats made available only to certain customer segments/tier levels? Yes. That's what AF already does where it restricts access to Premiere awards to (I believe) only Golds and Plats. There may be availability, but not everyone can get it.
The key difference is, that even if you leave availability open for too long, you might easily end up with lower income due to lower fares getting sold, but you can still zero out all classes at capacity (or capacity +1 or capacity +2, as you desire, whether that is specific flight capacity, capacity for the rest of the day, group capacity etc.). General guarantee mechanism (i.e. open to all eligible customers at all times) removes this option as it forces theoretically up to unrestricted (but always non-zero) availability.
Eventually, someone will use this non-zero availability to force an (undesirable) overbooking (which is not much of a problem in Y, but is a much bigger issue in J).
Eventually, someone will use this non-zero availability to force an (undesirable) overbooking (which is not much of a problem in Y, but is a much bigger issue in J).
Because I believe that booking guarantee demand is much more volatile than general demand and as such less predictable.
I would say the reason is necessarily increased volatility of guaranteed flight demand + inability to zero out a flight in advance.
I would say the reason is necessarily increased volatility of guaranteed flight demand + inability to zero out a flight in advance.
As we both agree, accuracy of predictions is key to maximise the yield, but the airline already has that information which they use today to make predictions and steer the availability of different fare buckets. What would change: operationalise it in such a way as to leave a nested inventory for those last minute bookers.
That is an entirely fair belief. Probably less likely to be shared by AFKL, judging by signals such as the state of Ultimate program...
I would still say it's riskier than not having such a scheme; that however does not mean it can not be worth it to an airline. I just think AFKL has decided, opposite to LHg (and in line with most other airlines around the world, it seems) that it isn't.
I would still say it's riskier than not having such a scheme; that however does not mean it can not be worth it to an airline. I just think AFKL has decided, opposite to LHg (and in line with most other airlines around the world, it seems) that it isn't.
Some? Certainly. Enough? That'd debatable.
But then, what is "enough"?
OK, maybe I shouldn't have used the term "whole discussion", although to be fair, only one other poster also mentioned short haul, in reaction to your reaction to them, and to point out that they don't disagree with you.
The difference is, if there is flight disruption in Amsterdam, you are down to 1/4 capacity. If there is a disruption in Frankfurt, LH still has 3/4 capacity. Or maybe 1/2 of F seats, if we take a hypothetical destination served from FRA and ZRH with F and MUC and VIE with no F.
It removes the concept of "sold out" at any time sooner than 24h before departure. And as you earlier yourself noted, there is not much sales in this period anyway...
In the end it comes back to the same principle: airline closes lower booking classes at a point in time to always have enough seats left in the highest booking class for the number of people that it predicts to take that highest booking class (BTW the video link I posted explains that really well). All I am suggesting is to make that highest booking class one that is only accessible to elites, only when all other classes are zeroed out, and only up to 24/36 hours before departure. Is it thinkable to have certain booking classes or seats made available only to certain customer segments/tier levels? Yes. That's what AF already does where it restricts access to Premiere awards to (I believe) only Golds and Plats. There may be availability, but not everyone can get it.