If you’re wondering what happened at Southwest Airlines (SWA) this week, Christmas holiday 2022, it all starts with a 1990’s program called SkySolver and another called Crew Web Access (CWA) that SWA uses to manage crew, routes and other operational logistics. 

SWA operates 4,000 daily flights, with an average of a 5-6 person flight crew, and many supporting ground roles. Each of their 700 planes can carry 150-180 passengers to its 121 destinations. The logistics of managing all those moving parts is so complex it requires software, artificial intelligence and critical planning to operate efficiently.

SWA schedules their plane routes by a “point-to-point” method, versus a “hub & spoke” like practically every other airline of a similar size in America. If one point fails, the entire system is at risk of collapse unless the software can keep up, and of course crew availability etc. but all managed by the software.

Let’s say the schedule is Philly, to Denver, to Dallas, then Vegas. This is point-to-point. But the crew gets stuck in Denver. Now Dallas and Vegas are missing both plane and crew and tomorrow mornings flight that starts in Vegas is already at risk.  That’s potentially 3+ flights, 500+ passengers and at least 1 crew that need rescheduling. Conversely, “hub & spoke” would have everyone flying to Dallas, then Dallas to destination. This makes it easier to manage risk. 

Other airlines have several of these “hubs” for just that reason. You have an issue, get them to the hub and problem solved, albeit with a manageable delay.

The software that needs to manage this, SkySolver, is old… ancient by any standard. Imagine telling your clients you run your operation on Windows 3.1 (1992). This is not an exaggeration - that is equivalent to what SWA is using to manage its operations.

SkySolver and CWA can handle ~300 simultaneous schedule changes at a time, anything beyond that it crashes. This is a very low threshold by todays standards. When this software was created in the 1990s, this was based on the maximum computing power at the time. And as hardware became more complex and computing power increased, SkySolver nor CWA can take advantage of this, and routinely crashed under even minor disruptive events. Nor was SWA willing to upgrade to newer technologies - they stuck with what they built. And yes, they built both of those.

Denver is a major way point for SWA and it was the catalyst for what was about to happen. 

As the storm got worse, workers called out. Staff was already thin considering the surge of call-outs in a virus heavy winter season. SWA sent a letter/email to all employees basically saying, “you’re all on mandatory overtime.” 

Mandatory overtime isn’t unusual in the airline industry. That is until the tone changed in the letter with this threat, paraphrasing, “No callouts without a doctors note upon return or you’re terminated.” And the threat of termination was repeated several times. It is said that upwards of 200 ramp agents in Denver walked out, quit or no-showed at the time the letter was sent. While this is not confirmed, former employees are validating this part of the story. See a copy of letter below. 

The letter itself is disturbing, inappropriate and goes against SWA’s own “people first” culture. Anyone would read this and be immediately uninspired and some, downright angry. Was this written by the same company with the core values of "Warrior Spirit" and "Servants Heart?" 

Even without the alleged walk-out, the meltdown would have happened, it just accelerated things.

No ramp agents = no plane parking or unloading. Planes stacked up as the storm intensified. 

With planes not able to deplane or depart Denver, other cities began to falter. Nashville, then Midway, then Dallas, Philly, Sacramento - even parts of the country with good weather at the time - failed. Planes started stacking up all over America. Unable to get repositioned, crews reached their max flight time, which cannot exceed 14 hours before mandatory rest periods.

The software was now being asked to manage tens of thousands of simultaneous changes. Just to manage a single flight persons schedule, there are 6 scheduling rules alone that focus on time. Other rules focus on qualifications, training and so on. It is hyper-complex for one person, let alone 20,000 people all at the same time.

They flipped to manually trying to calculate crew availability which is humanly impossible at the scale of SWA’s operation. It takes very powerful software and hardware. But they tried. Flight staff were asked to call their dispatch (or wait for a call) to check in and provide updated availability and get their assignments. There are many stories of crew members being on hold for hours just to get this information. This obviously had little effect on staving off the impending disaster. Unable to communicate effectively, and with crews  so wildly “out of position”, total collapse was now certain. 

Another point of failure is not commonly known. SWA is not connected to other airlines or 3rd party systems for rebooking passengers, repositioning crew, or even crew or passenger accommodations. All the majors and vendors talk to each other. Not SWA. You’re on your own. Even flight crews had to re-book their own travel and hotel.

There was a similar mini-meltdown in October ‘22 with SWA blaming ATC and weather in Florida. While those were contributing factors, that was not the root cause. It was the outdated SkySolver and CWA software crashing. Through some heroic efforts by employees, they were able to isolate it quickly and prevent it from spreading like a virus. They were once again, lucky, and this allowed SWA to feel a sense of false security.

We now know, through employee reports, that this has happened dozens of times since the early 2000’s as SWA has rapidly expanded service. Too many people, planes and places for the software to keep up.

SWA’s previous CEO, Gary Kelly, did not endear himself to the employees. He is an accountant by trade, not a “hands-on” operations person, as the founder Herb Kelleher was known to be. And Herb was loved deeply for his ability to roll up his sleeves and get dirty at every level. Gary, while competent, was focused more on financials not the frontline. SWA performed moderately on Wall Street… but was teetering on massive disruption with every major weather and/or other disruptive event. Gary failed to act and employees hold him accountable.

The new CEO Bob Jordan is an operations specialist. Given time, he has the ability and skills to fix this mess - if he listens to his frontline staff, among others.

Unlike other airlines, SWA showed an inability to recover quickly and efficiently because of:

1. A failure by leadership to follow their own core values
2. Old systems and software
3. Archaic routings prone to risk
4. Angry and disengaged employees who warned execs this would happen
5. Disconnected and disinterested leadership
6. Inability to work in the airline network for rebooking passengers
7. Hero Syndrome "Our employees will save the day"
8. Mother Nature

By the time this was all over, 16,000 flights were cancelled affecting more than 2.5 million travelers.

All this was a true perfect storm. All this made possible by lack of investment in better software. All this as the first domino, SkySolver and CWA together fell. All this as the employees at SWA deserved better as did the flying public. 

All this because leaders forgot their heart.