At American’s Investor Day two months ago, it started to create a justification for its focus on smaller cities using “network superiority” metrics. I wasn’t satisfied with those metrics, so in partnership with Raymond James, I built my own model. Now, it’s been a couple of weeks since presenting that to Raymond James’s clients, and I’m able to share parts of it with you here.
In its presentation, American said that it had network superiority in 59 percent of its markets in 2019, and that had risen to 68 percent by 2024. The data used to come to this metric was cited as:
via Cirium, March 2024; Continental U.S. + short-haul Latin America spokes where American offers most peak day departures and/or most nonstop routes among network carriers
There was a lot I didn’t like about this metric.
- It ignores Southwest entirely. In many mid-size markets, Southwest may not have the international reach but it has a far superior schedule with more nonstop destinations, so it absolutely should be in the conversation.
- It included short-haul Latin America cities. In many of those markets it doesn’t matter. The key in beach-type markets like many of those is which airline dominates in the US origin anyway.
- It didn’t look at the number of flights to each destination, just a total number from each origin.
- It didn’t include hub directionality — meaning American might service LaGuardia, Philly, and Washington/National but that wouldn’t necessarily make it better than Delta serving Salt Lake and Atlanta, because American would really serve only one “direction.”
- This ignores seat capacity entirely.
- It doesn’t look at actual connecting itineraries and destinations beyond the hubs.
All of this could be done fairly easily, so that’s what I built, again using Cirium data just like American did.
I took a look at all non-hub markets in the Continental US (excluding NYC, LA Basin, Atlanta, Dallas, Chicago, Miami, Washington, Denver, SF Bay Area, Houston, Charlotte, Seattle, Phoenix, Boston, Minneapolis, Detroit, Philadelphia, Salt Lake City) that were served by American in either 2019 or 2023. Then I looked at service levels on March 11, which was a Monday in both 2024 and 2019. And what I did I find?
% of Markets with American Superiority
The middle two basically apply American’s methodology to our dataset. The one on the far left uses our own methodology. And you can see that not only does this show American having superiority in far fewer markets, but it shows the gain between 2019 and 2024 to be far less as well.
Keep in mind that this is still an impressively high number. Out of the four big airlines, American has superiority in nearly half the non-hub markets it serves. That is more than anyone else. If you draw that line from New York to LA that I like to talk about, most of those American-superior markets fall south. United has more to the north. And Delta? Well, it just doesn’t have superiority in many places at all.
Markets Where Delta Has Network Superiority – March 2024
This means one of two things (or possibly both). On the one hand, it could mean that Delta just cares about those big, giant hubs where it does have superiority. That is, after all, where the lion’s share of the revenue falls. On the other hand, it could be that Delta just finds network superiority to be less important and can win on other attributes. Either way, Delta knows its market. Really, everyone does. I’ll show you what I mean.
I took an alternate metric to look at superiority in the hubs. Those are the mega-cities where nearly 80 percent of domestic revenue starts or ends. The airlines all have their own hubs where they have superiority, but this is where Delta and United really make their money. After that, I ranked cities by their 2023 domestic revenue.
Network Superiority by Airline by Market Size – March 2024
In the next tier down, that’s where Southwest shines. About 46 percent of total domestic revenues touch those markets, and Southwest owns them. It has also stretched into that next tier down, the 51-100, where about 17 percent of revenue touches. (Note, the revenue share doesn’t add to 100, because revenue can go from a hub city to a small city and count twice.)
It’s in that tier where American really starts to show up, and it continues all the way through the top 200 before we get into janky essential air service markets for the really small guys.
The issue here is that revenue falls off rapidly. If you look at everything from 101 on up, it adds up to less than 10 percent of revenue. American may have superiority in a lot of markets, but they aren’t big ones.
Let me put this in context. American has network superiority in 124 non-hub markets. All of those added up to just shy of $27 billion in revenue in 2023.
But look at Southwest. It’s fair to exclude Southwest’s top 5, because those behave like hubs: Las Vegas, Orlando, Austin, Nashville, and San Diego. But even just looking at the bottom 30 for Southwest, there is more total domestic revenue to be found than in American’s 124.
2023 Total Domestic Revenue for Markets Where Each Airline is Superior
I understand American wants to use its pilot scope clause advantage to flood regional jets on high frequencies in smaller markets to help get more high dollar traffic. The hard part is identifying where exactly that will work.
Obviously American’s fortress hubs aren’t changing. But if you go into markets that are too small, there isn’t enough opportunity to make those regional jets profitable. The key is really to try to go into that 51-100 range which is below the main focus for Delta and United and begins to be markets where Southwest won’t compete.
I’m not sure there’s enough here for American, but I can absolutely understand the desire to try. If you don’t want to (or can’t) compete with the others, then this is the best place to try and make a living.
43 comments on “Putting American’s Network Superiority Claims to the Test”
No surprise, really. Management talks its book, after torturing the data to suit its messaging.
“Out of the four big airlines, American has superiority in nearly half the non-hub markets it serves. That is more than anyone else. If you draw that line from New York to LA that I like to talk about, most of those American-superior markets fall south. United has more to the north. And Delta? Well, it just doesn’t have superiority in many places at all.”
You mean Delta isn’t the worlds perfect airline? A certain someone will be posting 3… 2… 1.
As for the NYC/ LA line, that’s not a surprise as American is in Dallas while United is in Chicago. Those second tier cities that Southwest dominate are quite formidable in their own right such as Las Vegas & Baltimore.
The NYC-LA line is interesting. Both UA (IAH) and DL (ATL) has only one hub south of the line and AA (ORD) has only one hub north of the line, within con-US, excluding focus city/partner/wanna be hub.
Being so regional heavy was disastrous for United Airlines, which is why Kirby had to spend billions to rectify the issue.
The casm of these smaller planes are horrid.
With such high costs you need to pull in high revenues to offset that and even if AA can find those markets, which is debatable, they would have spent so much time and effort on this that I seriously can’t see it being worth it.
I also want to point out that this is somewhat ironic. The airline infamous for the bean counters and keeping costs so low that they impact the passenger experience is totally fine with these high casm regionals.
It’s also the same issue that impacted USAir, especially prior to the AWA merger. One would think that lessons would be learned.
However, given the high concentration of AA hubs (5) up and down the eastern seaboard combined with heavier concentrations of population in this region, I think they fall into the trap of believing they can make it work.
Curious what are the one 101-150 and one 151-200 markets that Southwest dominates? it seems strange that Southwest would dominate such small markets with their only large planes.
Subway Nut – Well, the 151-200 is going away. That was Bellingham at 156 which Southwest is leaving. Then 101-150, however, is Islip at 105, so that isn’t going anywhere. And in case you’re wondering, the next lowest one for Southwest is Harlingen at 100.
I was really surprised by the statistic that 80% of domestic travel starts or ends at a hub.
For the analysis that you are doing, though, the real question is: *given* that travel starts or ends at a secondary city (The ones that AA is talking about here), what fraction of THOSE journeys have a hub on the other end?
For people who live in secondary cities, AA’s superiority statistic is irrelevant on any journey where you are not traveling to an airline’s hub. I suspect that this is more than 20% of domestic trips to/from spokes, but of course I’m not sure.
grichard – It’s a good question. In the 51-100 market size for FY 2023, 52.3% of domestic originating revenue went to a hub. In 101-150, it was 50.0%. In 151-200, it was 47.4%. And in the 201+ category, it was 51.5%.
So basically, about half the originating revenue in these spokes went to a hub. That would certainly skew which airlines had a chance at capturing traffic, it all depends on which hub they’re going to.
It looks like most of the Delta dominant markets in Michigan, Minnesota and Wisconsin are EAS markets.
America West does it again!
Chasing the once a year price sensitive traveler while stepping on rakes at every other opportunity.
The HP management group is doing their best to turn AA into America West.
Ed Beauvais rises from the dead!
America West does it again!
Chasing the once a year price sensitive traveler while stepping on rakes at every other opportunity.
Will everyone please stop with the false “America West did this”
narrative? The person in charge of commercial strategy who appears to have complete control is Vasu Raja, legacy American employee since 2004 who had nothing to do with America West. The VP of Revenue Management under him, Scott Chandler, has been at AA since 1996. The SVP of Network didn’t join AA until 2020, coming from WestJet and United before that. And the SVP of Partnership Strategy only came to AA in 2022 after JetBlue and United.
There is no America West leadership behind the commercial strategy other then CEO Robert Isom who seems to have given Vasu the keys to the kingdom in this area.
Thank you. I was going to say the exact same thing.
It is amazing how people create inaccurate narratives based on their false perceptions of reality.
Exactly. AWA was the only post deregulation carrier to make it. The AA old school folks love to whine and try to blame AWA for everything. Typical baloney.
Cranky – does your methodology overcome all of the reservations that you bulletted? How did you overcome measuring connecting itineraries?
AA’s methodology is laughably simple, and some other important metrics that it excludes:
– it doesn’t weight service in a market by the market size.
– it doesn’t measure the quality of the connections beyond the hubs. Where are the connections to? How much demand is on that o/d? What is the elapsed time of that itinerary, and how does that compare to competitors?
– it doesn’t take into account the timing of flights. 7:00 am departure is more marketable than 5:00 am. Daytime is likely better than redeye. In a round-trip, origin markets do better with departures in the morning and arrivals in the evening, and destination markets do better with midday arrivals and departures.
– It ignores how much their own service competes with itself. For example, one carrier may have more frequencies, but another carrier may have a better pattern of service with frequencies at different times. Also, for connecting itineraries, having options for highly varied departure times is preferable to having many options that depart around the same time.
– It likely overemphasizes service to places that once-per-year travellers frequent. FF loyalty to those passengers is less important on the path to maximizing yield. These are destinations like LAS, MCO, certain short haul Latin markets, amongst others.
– It only focuses on peak day schedules. AA really creates craters in their schedules on off-peak days, or outside of the peak season, through the process they use to remove capacity in off-peak times.
– It ignores the fare paid on the markets. Creating good service to a high fare market is more important than creating service to a low fare market, all else equal. DL and UA get some wild international fare premiums on o&d’s that only their respective alliances can create.
A well developed QSI (quality of service) model would overcome this. I find it hard to believe that AA, with the size of its commercial team, doesn’t have robust modeling capabilities that can produce higher quality measurements of the strength of their network.
Bobby – Yes, it does overcome all the concerns I had, though it could be refined somewhat to get more accurate. I don’t expect it would change the outcome. To answer your questions.
> – it doesn’t weight service in a market by the market size.
I think that’s by design, because AA thinks saying it has superiority in x number of markets makes it sound so much better. Of course, I did break it down by market size as I showed.
> – it doesn’t measure the quality of the connections beyond the hubs.
Where are the connections to? How much demand is on that o/d? What is the elapsed time of that itinerary, and how does that compare to competitors?
This is where it could be refined. I looked at domestic connecting itineraries with a 30 min to 4 hour connection time. OD demand wasn’t my concern. It was more about market presence in each spoke. This doesn’t get so deep into the weeds of a QSI model, but that’s because I was trying to bridge the gap between what American published and a more general but useful number than what QSI would provide. Since AA was looking at a single date, I wanted to take a similar snapshot. But schedules can vary enough by date and season that I had to create a model to keep it more generalized.
>– it doesn’t take into account the timing of flights. 7:00 am departure is more marketable than 5:00 am. Daytime is likely better than redeye. In a round-trip, origin markets do better with departures in the morning and arrivals in the evening, and destination markets do better with midday arrivals and departures.
It does not, but it does take frequency into account which is a good proxy. If there are 6 flights a day, then it’s going to have well-timed options. If there is 1, it very well may not. Time of day could be added to the model, but I don’t expect it would appreciably change the outcome.
>– It ignores how much their own service competes with itself. For example, one carrier may have more frequencies, but another carrier may have a better pattern of service with frequencies at different times. Also, for connecting itineraries, having options for highly varied departure times is preferable to having many options that depart around the same time.
Sure, that can be the case, but again it uses frequency to help change that. I added a “hub direction” to the model which lumps flights going in the same direction into one, like LaGuardia, Philly, and DCA. So it tries to avoid a fair bit of overlap. I was also pretty conservative on circuity to make sure it didn’t take into account hubs that would add absurdly-significant duration to the trip.
>– It likely overemphasizes service to places that once-per-year travellers frequent. FF loyalty to those passengers is less important on the path to maximizing yield. These are destinations like LAS, MCO, certain short haul Latin markets, amongst others.
Again, the goal was trying to do a better job of proving what American was trying to do. This isn’t about making a perfect model for a single day but rather a more general model that could give a sense of what’s going on in any given spoke. No question that frequency and spread throughout the day to a place like LAS is more important than, say, Springfield… unless someone needs to go to Springfield. This is just about overall service pattern potential.
>– It only focuses on peak day schedules. AA really creates craters in their schedules on off-peak days, or outside of the peak season, through the process they use to remove capacity in off-peak times.
Correct, because that’s what American did. Off-peak doesn’t matter as much anyway since people during those times tend to be more flexible and focused on price.
You get the point.
Thanks, Cranky. The bullet points below the question that I wrote to you were meant as a criticism of AA’s simple methodology, not as a critique of the analysis that you created. It’s my belief that AA should be able to come up with a more robust calculation of its network strength with it vast pool of human capital. I think that a lot is being lost in their high level methodology.
Bobby – Oh sure, and I imagine that American does have more internal metrics. The thing is, this was all about finding the metrics that would help support the story. I have to imagine that the airline decided this was the story it wanted to tell and then it crafted the numbers to back it up. Maybe that’s not right, but that’s how it looks.
I once did a similar analysis for the Dutch market. KLM was similarly as dominant.
Evaluating the Dutch domestic air market must have been… straightforward :)
Don’t underestimate Aruba Airlines
once again, another great analysis that can stimulate great thinking and conversation among those that understand the data and how it is presented.
Exclusion of WN from AA’s data and calculations is notable; WN does not do near as well in the largest markets but carries huge amounts of traffic in the middle tier of markets.
As for DL, it is clear that they shifted their network growth over the past 3 years of covid recovery to oastal markets where it saw a once-in-a-lifetime opportunity to grow away from its middle-of-the-US dominance. Coastal markets are much more capable of supporting mainline aircraft so it shifted mainline capacity from its core 4 interior US hubs. At the same time, low availability of regional jet pilots meant that RJs were not available per pre-covid CPA contracts for all airlines; DTW, MSP and SLC were all heavy RJ hubs and were medium sized hubs as well which meant that backfilling capacity with mainline aircraft resulted in fewer flights on larger aircraft = the opposite of schedule superiority.
United is in the midst of its NEXT program and will roll it out as soon as Boeing can deliver MAXs and based on whatever A321NEOs that Airbus can find for UA, both of which will allow UA greater penetration in small to medium sized markets. Schedule analysis like this equates a CRJ the same as an A321NEO but UA’s share gap will close as they get more mainline aircraft – which is a threat to AA’s focus on regional jets to serve small and medium-sized markets.
and the biggest caveat is that airlines ultimately are interested in revenues, not schedules. DL and UA particularly build their schedules to maximize network revenue which includes feeding international flights. Even domestically, DL still gets more revenue per seat mile than any other US airline so they target revenue maximization rather than network “bulk”
The bottom line is that this data is all very much in a state of change. As more RJs come back online for DL and UA (AA has paid more to restore its regional network) and as DL and UA’s networks become more alike – DL more international, UA more domestic – this analysis will look very different in a couple years.
As Mark Twain once observed, “There are three kinds of lies – lies, damned lies, and statistics.”
In quoting Twain, it’s my my intent to call anyone out. But it is my intent to point out that the same data can be interpreted differently, and those interpretations all have some validity. The key is to let those who consume the data know the methodology behind the numbers, so he or she can put things into proper perspective. That’s if there is such a thing.
The point I want to emphasize is based on slide number 97 of the investor presentation. Why do I want to emphasize this slide? It’s because of an assumption I consistently read that American wants to massively grow its regional fleet. The slide states otherwise. It shows American’s projected regional fleet maxing out at roughly 580 aircraft – not growing. If American’s narrowbody fleet gets to 900 (which the slide projects), the number of total regional aircraft allowed under its scope clause would be 675 (75%), and the number of 76 seat aircraft allowed would be 360 (40%). Maybe my math isn’t what it should be, but 580 is 95 fewer aircraft than could potentially be allowed. American currently has about 840 narrowbody aircraft. Under its scope clause that would mean American can have 630 total regional aircraft and 336 76 seaters. According to its March 2024 update, the airline currently has 556 total regional aircraft and 284 76 seaters. That means American is currently 52 aircraft short of the number of 76 seat aircraft it’s currently allowed under its scope clause and 76 short of what would be allowed once its narrowbody fleet grows to 900.
Looking at the map of where Delta’s network has superiority you notice a huge hole in there network from the mid-Atlantic through the southern Midwest, the Plains, parts of the South including Texas and a huge swath of Mountain region. I don’t want the Delta faithful to attack me but I have to wonder with a hole that big in their network where they have no dominance how are they making money?
They must be doing something right since they are the most profitable airline and have been for a while now.
Delta has NYC, and the Detroit and Atlanta hubs work very well at getting me where I want to go quickly (BWI is my main airport, but I don’t like Southwest). PHL is closer by car, but inconvenient by train, and it and American suck. Similarly, from RDU, Atlanta makes good sense as a hub if you’re staying South. Detroit or SLC for other purposes.
Is this how the major hubs break down in terms of domination?
AA: Dallas, Charlotte, Phoenix, Philadelphia, Miami
United: Chicago, Denver, SF Bay Area, Houston, Washington
Delta: Atlanta, LA Basin, Seattle, Boston, NYC, Detroit, Minneapolis, SLC
VictorKilo – Yep, you got it. But I do hate to call it domination, because some are closer than others. Like in LA, United is right behind Delta and American is right behind United. In New York, United is on Delta?s heels with others far behind.
I agree, it’s close enough in several markets that I had to ask the question.
https://www.panynj.gov/content/dam/airports/statistics/statistics-general-info/monthly-2024/REG_MAR_2024.pdf
Cranky, just curious which metric has DL bigger than UA in NYC. I agree it’s very close and neither is “dominant”, but the rolling figures show UA carries more pax in NYC.
Mark – That’s the whole point of the model I built. It’s not just about passenger numbers. It’s about utility to the region.
Yes, but you forgot United has a huge hub at Newark & therefore they control a sizeable swath of the NYC market as well as international service/ feed.
In your first figure, what is the difference between the “AA Methodology Absolute” and “AA Methodology Equal or Better”? Just a matter of > versus >=?
Angetenar – Yeah, exactly. In “absolute” it has to be > in at least one of the two metrics whereas in the other one it can be =>.
Great analysis.
Lots of complex questions and with all the time an analysis in the world – i would look at each market and for each O&D pair in the market run a combination of number of flights within specific arrival and departure time buckets (do i have a morning, afternoon or evening option) and weight by travel time. Basically trying to ask the question – if I’m trying to go from X to Y – does American have the best option (time of day and total travel time). If i could weight it by revenue spend even better. For example – Manchester, NH (MHT)- a market that southwest dominates to Orlando but can’t get you to London but American can. Delta doesn’t serve it. United currently can’t get you back from London without an overnight in Newark. The question is how much O&D is there from MHT to Europe and other other destinations that American’s RJs to PHL, DCA and CLT create a network advantage at $2,000/Y ticket RT vs. MCO where Southwest focuses on (where you have more seats but a lower average fare)
Thinking about this matrix of Network Superiority – American still looks like the best across network superiority. Would you trade AA’s network position with DL or UA? Sure AA is behind DL on many large market but it’s certainly not a winner take all. In those large markets, the market segmentation will depend on the destination. In NYC – if you need to fly to finance hubs: ORD, LHR, LAX, TYO, AA covers you pretty well despite not having a flight to ATL (pretty big network gap).
I used Fargo before as an example of an airport where it’s very difficult to say one airline has a “superior network.” AA claims its 2 DFW flights, 3 ORD flights, and seasonal PHX flight are superior.
However UA has more flights with 3 each to ORD and DEN, but most of these are 50 seaters. DL has 5 flights to MSP for superior time of day coverage with 2 of these on mainline equipment. And don’t forget Allegiant, who serves 6 destinations, although only 1 of which is daily.
Who really has the superior network? It really depends on the type of customer.
Bravenav – If you’re wondering, my model whos American has having superiority in Fargo, followed by United, then Delta. This is March, so the Phoenix flight certainly helps. Delta does win on frequency in terms of how many times it can get people to various cities around the US. And United wins on total number of destinations you can reach via connections.
@Cranky How do you not consider BWI and DAL (and maybe DEN) not “hubs” for Southwest? Southwest has a more distributive network than the big three, but I would lump those in as hubs where it takes connecting passangers.
Brian W – Those are all considered hub markets in this model. BWI falls into the broader WAS market while DAL is part of DFW/DAL.
As someone who has competed against Delta, you can’t help but tip your cap to the first map in the article.
It’s amazing how few markets which show Delta having network superiority yet they are the most profitable US airline.
They clearly have a playbook that allows them to compete and be profitable to a degree that no other airline has done.
@Cranky This is the post that got Vasu fired. Your fact based, methodical decomposition of AA’s network claims surely made its way into the thinking of the mainstream media and the analyst community. A lot of the current press is about their flawed distribution strategy. The problem is more in their network. They retired a bunch of widebodies during COVID while taking victory laps for cost control. When int’l demand came back, they didnt have the airplanes to service it. UA kept their options open. Ultimately, the decision to fire Kirby has cost shareholders dearly. Long Kirby, short Parker/Isom is an enduring trade.