Spirit Focuses on Its Schedule to Improve Operations

Spirit did not have a good summer in 2021.

For years, this wouldn’t have been a big deal. After all, the airline prided itself on only caring about offering low fares. A terrible operation was just part of the package. But the airline changed, especially when Bob Fornaro came in as CEO in 2016, followed by Ted Christie in 2019. The operation became important, and the summer of 2021 proved to be a low point as the airline raced to capture the quick return of demand post-COVID.

Spirit Airlines Operational Performance By Month

Instead of just deciding to live with a poor operation, Spirit decided to take action. The airline did the usual things, like adding more crew buffer in more of its stations to help deal with increased absences or delays, but it also went beyond that.

Spirit switched its aircraft routing model away from a Southwest-style point-to-point system to an out-and-back model. Instead of going, say, LA to Vegas to Dallas/Fort Worth to Fort Lauderdale, it would run an airplane from LA to Vegas and back, then from DFW to Vegas and back, and then from Fort Lauderdale to Vegas and back (or something like that). Since Spirit sold very little connecting flights away, there wasn’t really much appeal to the point-to-point model for the airline in the same way that it appeals to Southwest.

Spirit was able to create this new schedule beginning in April 2022, and I’m told it did not see an impact on aircraft utilization.

The airline also thought about how it brought down its off-peak flying on Tuesdays and Wednesdays. Instead of doing deep off-peak cuts, Spirit decided to go with more stability since it made it easier to schedule crews through the week.

All of this combined allowed the airline to post a much better summer in 2022 with minimal cancellations. But wait, there’s more.

Just last month, Spirit announced that it had purchased Amadeus SkySYM, software that would help the airline optimize its schedules using artificial intelligence (AI). It’s a part of a system that Southwest has been using since 2018 which was originally built by Optym and sold to Amadeus. Any time I hear about an implementation of AI, I assume it’s just a buzzword, so I reached out to Spirit to learn more and had the chance to speak with John Kirby, VP of Network Planning.

The idea beyind SkySYM is that it allows you to feed history into the system, so that it can then analyze future schedules and suggest ways to improve them. It’s a form of machine learning, as I understand it.

To understand whether this would really work, Spirit had to see how well the system could predict actual performance. It loaded in historical schedules and had it predict performance. Then Spirit could compare that to actual performance and see if it was worthwhile. While the system naturally can’t accurately predict when thunderstorms will sit on top of New York City or decide to put Fort Lauderdale underwater, it did predict regular performance “very closely” and hasn’t given any really strange results.

With history loaded into the system, Spirit can now move ahead with the next part of the plan. Starting with the July schedule of this year, Spirit has loaded the new schedule into SkySYM and run it in parallel with its normal schedule planning process. If the changes make a lot of sense, Spirit will incorporate them into the system. If there’s something wacky in there, it’ll give Spirit pause. The airline will then try to pick apart why something is being suggested so it can feel fully comfortable.

I asked John to explain the scope of this system. Is it thousands of changes? And how broad is it thinking? Will it suggest canceling flights, for example?

John says it’s probably going to suggest somewhere “in the hundreds in terms of things you can do to improve the operation.” And this isn’t about wholesale changes to schedules. It’s more about adding a little block time on one flight, or reducing it somewhere else to reallocate it. It may suggest adding 5 minutes to the turn time in one place based upon the model it has built.

In the end, these hundreds of things do add up to something. John said the airline thinks this can improve on-time performance by 1 to 3 points. That may not sound like a lot, but with 25,000 monthly flights that means 250 to 750 flights that no longer have an issue. That’s a real impact.

Spirit is already feeling confident that this system will work well after all the testing it has done. Once it puts it into actual production for a couple months, it will start trusting the system more to make these small changes so that it can keep trying to improve the operation, just the next step in the plan is put in place two years ago.