Suites don't win our business on merit

I've given the same advice at dinners in Houston, Dallas, San Francisco, London and Montreal. Different companies, different products, same advice.

The question at the end is always some version of the same one. What would make us matter more to airlines?

My answer is never the one anybody in the room is hoping for. Help your customers become data independent. Including from you.

It sounds like advice against interest. It's the opposite.

Every airline has lived some version of this.

The airline decides to get serious about fuel optimization. Goes out and hires someone expensive from FedEx, someone who has actually done the thing. Nine months later that person is still assembling data. Not modeling. Not optimizing. Assembling.

Swap fuel for revenue management and the story doesn't change. Swap it for loyalty, or network planning, or anything else we've decided to get serious about this year. The expensive hire arrives, finds the inputs scattered across systems nobody owns end to end, and spends two quarters doing the data engineering that's in someone else's job description.

We read that as a hiring miss or a vendor that underdelivered. It's neither. Airlines don't have a talent problem, and most of the time we don't have a software problem. We have a data access problem that makes the talent and the software both look like they underperformed.

And that's the part that quietly decides what we buy next.

Sit in the seat of an airline that just lived through that. The next time a real decision comes up, integration burden isn't one factor in the evaluation. It's the factor. Nobody says it that way and no scorecard has a column for it, but it's in the room.

So the airline buys the suite. Everything in one place, one number to call, and the promise that this time it will be easier.

Not because the science is better. Because it's easier—and easier is what you're buying after nine months of watching an expensive person assemble spreadsheets.

Which means the software with genuinely better math and genuinely better models just lost a deal on something that had nothing to do with better. And never found out why, because the question wasn't what's better but what's easiest.

I've watched the other version enough times to trust it. At three of my largest engagements, within a year to 18 months of us building a complete unified commercial data layer (bookings + tickets + revenue accounting + inventory + forecast + loyalty + DDS + competitive fares), each of those airlines put out an RFP and chose a science-based, independent revenue management system over the engine they'd been running on faith for years.

That was never the goal of our work. It's just what rational commercial organizations do once they can see.

Airlines in control of their own data didn't cost the independent software vendor any business. It's actually what let them compete for it.

Then there's the question none of us answers well.

I tell clients they need their own data for two reasons. One is to run the business. The other is to know whether the vendors they're plugging in are worth what they cost. Because if the only view of value is the vendor's own forecast, the vendor is always going to tell you it's working.

That isn't an accusation. Lift is counterfactual—here's what happened, here's what would have happened without us—and the second half of that sentence doesn't exist anywhere. Somebody has to build it. When the customer can't, the vendor does, and hands over a number none of us can check.

I've watched honest vendors lose to that. People who model value carefully and quote an actual range, beaten by a competitor with better packaging and a bigger number. The fix isn't a better story. It's a customer who can check your work for themselves.

There's no perfect way to measure lift. There are decent ones. A before-and-after on a window you chose. Two like-kind markets where the system manages yield in one and doesn't in the other. Imperfect, arguable and yours—which is the only attribute that matters when the number has to survive a budget review or be counted on to deliver the team's bonuses.

Every piece of software an airline buys moves it toward data independence or away from it. There's no neutral option, and the vendor almost never volunteers independence.

So ask during the evaluation. Once this is running, can we get our own data out—at the lowest grain, on a schedule we set, without asking permission or tickets to a help desk? Can we compute your value claim ourselves?

The companies building something durable answer yes before you finish the sentence—it's already built into their product.

In a world where intelligence and translation are ubiquitous, the software companies that last in this industry won't be the ones that make themselves hardest to remove. They'll be the ones whose customers could prove they were worth keeping because they could leave with their data at any time.

Allegiant built its whole model on refusing to sell through anybody else. On July 14 it signed a 12-month exclusive with Expedia, its first authorized OTA, putting 566 routes across 124 U.S. cities into Expedia's brands.

Southwest ran this play in February 2025. Metasearch first, then Expedia and Priceline, the first OTA partnerships in its history. Within 12 months: bag fees, a basic fare and the end of 54 years of open seating. By February, Forbes was asking whether Southwest had become just another airline.

The Expedia deal didn't cause any of that. Both came from the same pressure. But distribution was the first move visible from outside, and it's the one that decides who gets to learn the customer. Allegiant may have no choice here. It's worth being clear-eyed that the price is owning less of the thing that made it different.

Distribution was one story this week—assembly was the other.

The FAA awarded Air Space Intelligence a 12-year, $875 million contract on June 22 to rebuild how flights get scheduled and routed across the national airspace. Palantir lost. Thales lost. That's the story everybody ran, and it's a good one. The better one is in the FAA's own problem statement: the inputs controllers work from—weather, airport capacity, the rest—live in separate systems, and controllers bridge the gap with extra screens and manual workarounds. So the first thing the largest air traffic modernization in generations actually does is put that data in one place. Not a new engine. Not a smarter model. Assembly. We have wanted this for decades, and when it finally got funded the work started exactly where it always starts.

Two Portuguese meals in one week, in two countries.

Provincetown first. Draw a line due east from the tip of the Cape and it lands on the Azores, which is roughly how the Portuguese got there—whaling ships left New England short-handed and stopped in the islands to pick up crew. By the early 1900s they were close to half the town.

Then Montreal on Wednesday, in the Plateau. That community arrived a century later. I love how food is a constant across continents and centuries and how travel lets us discover that universal connection.

The wine list ran pages deep in Portuguese. I picked the Herdade do Portocarro, from Setúbal, because they grow Cabernet Sauvignon and Petit Verdot alongside Touriga Nacional and Aragonez and blend them together. Not quite a Portuguese wine, not a French one. That's my kind of systems integration.

This week I was told something couldn't be done because the system requires it a certain way. So I asked how the airport does it, since the airport does it every day. The answer was a manual workflow an agent runs by hand.

The limit was real. It just wasn't technical. A real constraint has no manual workaround. Manual is one tool. We have a lot of new tools. As leaders, we can't let habit turn into dogma.

—Ben