The Problem Must Come First

There is a version of the AI conversation happening inside airline commercial teams right now that begins with the solution. A team decides it needs a competitive pricing agent, or a revenue management co-pilot, and then works backward from there, figuring out the data, the workflow, the governance, and the success criteria. All of it serves a thing the team decided to build before it was clear what problem the thing was supposed to solve. Melissa Skluzacek, who has spent more than thirty years in revenue management, will be speaking at the World Aviation Festival in Lisbon this year about this pattern. She sees it as a common reason AI investments in commercial operations underperform.

“I think it fails because people can become focused on building AI instead of solving a problem.” Her example is one every commercial team knows: an airline notices it is reacting too slowly when a competitor changes capacity or pricing, and the instinct is to build an agent that monitors competitors. But the question that precedes that decision has not been asked: why is the reaction slow? Is it the analyst-to-flight ratio? The fragmentation of information across systems? The difficulty of identifying which competitor moves matter? Once the root cause is understood, the answer might involve AI, or it might involve better data pipelines, or a different team structure, or all three in some combination. “AI might be part of the answer. It might help monitor markets, summarize changes, and surface insights, but there could also be other things that are needed to improve, such as the data, the workflows, automation, or how the teams work together.”

The reframe she returns to is consistent: “AI is not the strategy; it’s a lever to solve a problem. The strategy is improving the commercial outcomes. AI is one of several tools that can help us get there.” This may not be the framing most attendees in Lisbon are anticipating when they walk into a panel on AI in revenue management. They are looking for the answer, not a methodology for finding the right question. Her advice is to flip it: “Think about what problem you’re trying to solve and how AI can help solve the problem as opposed to looking at it backwards and saying, what can AI do for me?”

Adoption and trust are, in her view, the same conversation. You cannot get one without the other, and neither comes from a big rollout. “If you start small and you really are sure it’s giving decisions that are right, like you could do it yourself and validate very easily, then you’ll know that, okay, I trust this. It is making the decision I would make. And then you just keep building on that over time.” The practical mechanics of this are less complicated than the organizational politics: pick a narrow use case with a visible and verifiable output, run it alongside what the team would have done without it, and let the comparison make the case. The confidence that builds from seeing a system perform correctly on something small is the foundation for handing it something larger.

Getting the team involved is an early requirement. A tool that analysts feel was designed around them, rather than handed down to them, starts from a different position when it comes to adoption. Melissa is emphatic that the best solutions in revenue management are built with the team, not for them, and this is especially true when the tool in question will be modifying how analysts spend their time. “AI is very scary for a lot of people… [If it’s] trying to replace them, or it’s making bad decisions, you’re not going to get there.” A lot of people in RM are genuinely uncertain about what their role looks like as automation expands, and a tool that feels like a threat will be resisted even when the logic for it is sound.

On the measurement question of how you know whether AI adoption is working, Melissa has a view that most organizations would rather not hear: the measurement problem predates AI, and AI is mostly exposing it. Airlines measure commercial outcomes well enough, but connecting individual revenue management actions to those outcomes is hard, because so many factors move the result at once. The right response is not to retroactively measure past performance but to start recording decisions going forward: what the hypothesis was, what action was taken, what happened afterward. That way the system accumulates a record worth learning from. An AI that can reflect your own decision history back to you is a fundamentally different kind of tool than one that produces recommendations in a vacuum.

Melissa’s comments echoed our recent series with Scott Davis, and converges on the same place, regardless of the starting point. “If data is your problem, obviously your AI isn’t going to be good if your data’s bad. Start with getting your data going and just continue to move forward.” She adds something that the conference circuit rarely says plainly: “I think a lot of people are trusting AI more and more. Maybe too much, to be honest.” Trust has to be built through validation, not assumed from a vendor claim or a successful demo.

A year ago, Melissa expected to start slowly, with things taking off a few years from now. Instead, “we’ve hit that hockey stick faster than I anticipated.” What she wants to see by next year’s festival is not “that we just threw everything in it and everything is AI,” but “that we were very thoughtful about it.” The conversations will have changed: when someone is talking about an opportunity in a market, they are connecting the dots across all the levers rather than optimizing in isolation. Start small and keep building, she says, then look back six months later: “you’ll be really surprised at how far you got.”

Karyn Fernandes spent two decades in pricing, revenue management, and system implementations at Sun Country Airlines and is also the author of “Minnesota’s Phoenix: From the Ashes of Braniff to Sun Country Airlines”, the definitive account of the carrier’s founding. She brings deep operational expertise and hands-on technology experience to product decisions, ensuring they’re grounded in real-world revenue management practice.

Karyn Fernandes spent two decades in pricing, revenue management, and system implementations at Sun Country Airlines and is also the author of “Minnesota’s Phoenix: From the Ashes of Braniff to Sun Country Airlines”, the definitive account of the carrier’s founding. She brings deep operational expertise and hands-on technology experience to product decisions, ensuring they’re grounded in real-world revenue management practice.