This week a chief commercial officer told me joining data was easier in 1995. He started his career at another legacy carrier about ten years before I got my first gig at Continental. He wasn't just reminiscing about the good old days. I remember the difference when I started, too.
One Set of Hands
In the late 2000s, I could pull revenue, booking and loyalty together in one shot. We even had a bit of the early digital and ancillary data coming online. I could give my VP answers in a couple of days. Sometimes even the same afternoon.
A day. Not a quarter. It wasn't a project. There were no tickets.
This commercial exec talked about his current team not having the ability to do the kind of reporting he did himself using QMF as an analyst all those years ago.
Decades of "progress" sit between his and my first aviation jobs and today. Storage became cheap. Compute became expensive. Everything got rented. We've bought tools the analysts of 1995 couldn't have pictured. And the simplest thing we do—getting two systems to agree on revenue by market—stopped being simple.
Sit with that. Not that the data got bigger or the queries got faster. That a plain question got harder to answer while everything around it got better.
Where It Went
So what happened? Not a disaster. A migration. Piece by piece, airlines moved the data out to a sprawl of platforms—loyalty to a system that counts more than flights, marketing to a suite that tracks digital journeys, the warehouse to whatever database the white shoe firms were pointing to.
Every move solved a real problem. Every move also put a little more distance between a question and the place answers lived.
None of this was intended to slow things down. The CIO inherited a stack three predecessors deep and a mandate to modernize it. The vendors shipped exactly what airlines signed on the dotted line for.
What eroded wasn't competence. It was proximity. The data didn't get worse. It got further away, and it stopped belonging to any one person who could stand behind it end to end.
In 1995 one person could hold the whole path from question to answer, because there was barely any path to hold. I could still hold it in the late 2000s for the same reason. Each new platform put another step between the question and the answer—adding product managers and request tickets along the way—until no single person could carry one all the way through.
Image: the future
So the data has to be reclaimed. And it's almost never reclaimed by buying something.
It gets reclaimed with leadership. I've seen it. It's usually commercial, usually the one who's out of patience—who asks why a basic question now takes a quarter and won't accept "it's complicated" as the answer. And is willing to tell people what good looks like when it comes to data-driven business.
That CCO on my call wasn't mourning 1995. He's calling it like it is, and he's the early sign another airline is about to take its data back.
What this means Monday morning
If you're building the data: Pick the report you hand off most. Do you know the decision it feeds, or just the spec you were handed? Spend an hour this week with the person who actually reads it. That's you shifting right—toward the problem instead of the pipeline.
If you're leading a data team: Draw your team on the line. Who's pointed left at ingestion, infra and security? Who's pointed right, into the business? If they're all bunched in the middle clearing pipeline tickets, you've got an availability gap on one side and a distance gap on the other. Move one engineer into a commercial standup this week and see what comes back.
If you're the executive funding the next investment: Take the question your team keeps answering "in a quarter." Ask who owns the path from it to the answer. If that's no one—or a queue—you've found what to fund. Not the next platform. The next dollar pulls infrastructure left and pushes your data people right, closer to the commercial problem.
In my feed
Two weeks, two acquisitions, one move. Salesforce bought Fin—the company that spent years insisting it defined the customer-agent category—for $3.6 billion. And MoEngage bought Aampe, the engine running 200 billion per-user decisions a week. Salesforce's answer to composable is money. But here's the thing: owning the layer was never about the software. It's about the data going in and out of it.
The loudest sell this month wasn't a deal—it was a report. Amadeus and Microsoft declared 2026 the year agentic AI goes from pilots to production. Maybe it is. But a vendor telling you the future arrived and you're already late for it is FOMO theater. Don't take the bait. The work doesn't change because someone published a deadline. Focus on the business. AI or no AI, do the next best thing with your data that drives an outcome. Then do the next one after that.
Speaking of big tech, contracts and money. If I hear one more time that an airline's signed a software deal with Sabre or Amadeus without explicitly defining the data that comes with it—down to the fields, records, latency and delivery method—I'm just going to give this whole thing up and go work for a bank buying and selling miles for a living.
Intelligence is going to zero. Software isn't the asset. The data that falls out of it is. And that's the line item nobody writes down. Define it in the contract, or you'll rent your own data back one query at a time.
What I'm cooking
The kids are deep into World Cup season, which around here means a sticker album, a growing pile of doubles and a standing table at the neighborhood pub. The album is mostly empty slots right now. The kids will trade with anyone who sits down, and the negotiations are more ruthless than anything I ran into this week about data.
The carry-on
No matter what we call the things we build, it's not the labels that create the trust. It's the outcomes. Positive outcomes build trust. And "change moves at the speed of trust."
—Ben