Monday, October 19, 2020

What's Southwest's Network Strategy?

 Note: Over the last year, I have intentionally stayed away from much direct analysis of Southwest Airlines. As discussed in my ethics pages, I am a former SWA network planner and was aware of the network strategies and tactics. However, after being away from the organization for two years and COVID-19 tossing all the network planning rules out the door, I'm confident to say I no longer have inside information, and I'm comfortable analyzing SWA's newest network changes. The discussion below is entirely based on publicly available data and my best thoughts on what is going on at the airline.   

COVID-19 has turned every airlines' revenue management, marketing, and network planning strategies completely on their heads. Gone are the good ole' days where ticketing revenue was predictable down to the day and hour with +/- .1% error. Legacy carriers have completely shifted away from their traditional 330 days of rolling inventory and are now making monthly adjustments to their entire network structure. 

ULCC carriers, which generally skew towards leisure demand, saw quicker recovery rates over the summer, just to see them tail off in the latter part of August into September. While 3Q TSA screening rates have "recovered" to 28% of the prior year, Delta and United have reported that ticket revenue was only 17% and 16%, respectively, of last year. While other revenue streams were "better," overall revenue was down roughly 80% for both carriers. While I suspect LCC/ULCC carriers to post slightly better results, better is a relative term. All airlines are hemorrhaging cash, and it's not just a flesh wound. 

The massive amounts of debt airlines are currently incurring to stay alive will limit their ability to grow by any significant measure for the foreseeable future. This can be seen by Boeing's latest forecast, where they do not expect the historic 4% growth rate to recover for at least five years. 

While I plan a deeper debt to growth analysis in the future, we can say with some certainty for each multiple of billions of debt an airline issues to survive, we will see a corresponding increase of deferred future growth. As the CARES Act expired, buyouts and layoffs at carriers kicked into high gear. Even Southwest is now issuing warnings to employees of the need for salary reductions or furloughs to reduce cash burn. None of these actions should be considered typical management vs. labor disputes/bluffs of the past. If carriers continue to issue debt to survive, we should expect to see five-plus years of limited growth, hub closures, and aging fleet.

While airlines try to reduce cash burn, they are also becoming increasingly nimble with their network. Both American and Delta have cut smaller cities following the CARES expiration to reduce costs and eliminate controllable cash burn. Southwest, however, is taking a different tack. Southwest has announced six new cities to tap into new revenue streams. For those not following airlines day-to-day, these new cities are Palm Springs, Houston (IAH), Chicago (ORD), Miami, Montrose (seasonal), and Steamboat Springs (seasonal). And while this set of new cities seems quite diverse, my goal today is to show you they all seem to follow the same approach. 

Before we get to the current network state, we need to look at Southwest before COVID. While we all like to rib Southwest for selling TPA-STL-DAL-PHX-LAX flights, a majority of the Southwest passengers fly on single coupon tickets. In other words, most of Southwest passengers fly nonstop or direct. Granted, over the years, this has been somewhat in decline. 

Southwest's network is incredibly complex and simple at the same time. While Southwest often states it doesn't have hubs, it does have ICOs or intentional connecting opportunities. Generally, ICOs are designed as small connecting banks, often 10 or fewer flights, strategically placed around the networkoften within the largest Southwest cities. While the ICOs allow structured connections for passengers to flow across the system, the Southwest network generally focuses on nonstop demand. However, in the COVID environment, Southwest's connecting network design appears to have been placed on steroids.

There are quite a few changes in the Southwest network that took me by surprise. First, schedule volatility has increased significantly. In 2019, schedules were usually only adjusted on Saturday and holidays. In 2020, it seems that each day has its own schedule. 

Let's take a deeper dive into an individual day. For my analysis below, I am using the December 7, 2020 schedule and comparing it to the week's same day last year (December 9, 2019). Really any day could show a similar trend; however, I decided to pick a day that appears to be a stable and repeated schedule for at least a couple of weeks. At a high level, we see SWA is much slower to ramp up the network and terminating at a much earlier time than 2019. This should be expected as demand for early morning flights and late-night flights is generally much lower than the day's core. 

Nothing surprising so far, but let's take a look at individual cities. When we review the largest SWA cities, we see that the network is under a complete transformation, enabling many of the new cities to come online. Let's take a look at MDW. In 2019, we can see hourly departures averaged between 10-15 departures with some peaks. Those peaks were the ICOs we previously discussed. ICOs are elevated times of departures but not typically heavily concentrated like we would see with American in CLT. However, when we look at 2020, we see the ICOs were removed for three consolidated peaks. In between the peaks, the Chicago operation is relatively idle. 

ATL's flight structure has more extreme banks than MDW. In 2019, ATL averaged around five departures an hour with small departing ICOs in the morning and evening. In 2020, if the flights are not directly connected from ATL's two large banks or a more shallow 3 pm bank, flights really don't operate. When we look at the flights operating outside of the banks, most of the flights depart for another banked mega city.

This structure is not limited to ATL and MDW. I found consolidated banks in ATL, BNA, BWI, DAL, DEN, HOU, LAS, MDW, OAK, PHX, and STL. I could even be convinced there is a mini-bank in SJC as well. But what does this all mean?  Well, let's take my home airport, OKC, as an example. Oklahoma City only connects into cities with banks, and most flights are feeding banks. 

When comparing the network year-over-year, 88% of Southwest flights are now connected to cities with banks. This is up year-over-year by 11pts. What is left within the Southwest network that is considered point-to-point? Calfornia, Hawaii, and Florida. Outside that, if the route is not connecting into one of Southwest's largest cities, it was eliminated. 

Let's take a look at another legacy small Southwest city, Tucson. In total, Tucson has had their flights reduced from 13 daily flights to five. When we take a look at the cities removed, non-banked cities were removed entirely. Chicago, a long haul flight, was also removed in place of Houston, which hits the banks as well. 

So, what's all this have to do with the new cities? Well, everything. Let's take a look at Palm Springs and Miami, which launch November 19th. PSP connects right into the banked cities. Out of the five flights operating into PSP on December 7th, four of them are directly connected to banks feeding the city. 

Miami follows a nearly identical approach to the design. For most cities, MIA is near or in most banks, excluding flights to TPA. Honestly, I can't figure exactly what SWA's thinking is with flights to TPA from MIA unless they are expecting organic connectivity on TPA's Midwest flights. 

As the new city list expands to Montrose and Steamboat Springs, the strategy largely stays the same with flights connecting into Denver's banks. I suspect O'Hare and Houston Intercontinental could take a similar approach.

With the extreme banking and new cities, does this mean there is a change on the horizon? I don't know about the long term, but I believe that we could continue to see Southwest expand their city list in the near term with the push to diversify revenue. 

If I'm reading the ground ops union contract correctly, Southwest only has to notify the union of their intent to outsource. There is no explicit requirement for Southwest to insource ground ops until a city operates more than 12 daily flights. No city, excluding Miami, is even close to this. This allows Southwest to spread their fixed costs and access new revenue streams without new employee costs. If Southwest employees handle larger cities such as Miami, O'Hare, or Houston Intercontinental, the company could allow bids from underutilized employees throughout the system. Again, this would come without much incremental cost to the company. 

This should raise questions for air service development professionals. If the negative impacts of COVID continue and airlines desperately look to diversify their revenue stream, could we see Southwest operate more new, small cities? Absolutely. With the Southwest network banked and seemingly relying on connections, it seems probable that we will see more cities announced by Southwest, not city closures that we have seen from other carriers.  

On the flip side, if you are an existing Southwest city that is already connected to the nearest banked cities, it is unlikely that you will see much growth unless you have demand for a point-to-point leisure location such as Florida or Hawaii. 

With earnings coming up on Thursday, we should hear some conversation around Southwest's future network strategy. It would not shock me in the slightest to hear additional future network plans, including routes for ORD and IAH, or if more cities may be added to the network. If we hear significant updates from Southwest's comments, I plan on updating this article. 


  1. In the last set of charts, why are flights at the beginning of banks indicative of "banked city connections"? All but the latest two HOU-MIA flights fall at the start of the bank structure... I would have expected them to be at the tail of bank structures if they wanted to capture connection traffic. But maybe I am thinking about it flipped?

    1. As long as the flight is in or near the bank it should get most, if not all, of the inbound connecting flights. What is not shown in the charts are the inbound feeding flights. Generally, you would see them arrive 45-60 minutes prior to the next outbound flight.

      Due to scheduling restricting throughout the day, you might not get a flight perfectly placed as you would like. That line might be feeding another bank in the system or tied to a slot time. So adjusting the flight to be perfectly timed might not be possible.