The fix was months away. The business couldn't wait a week.
A carrier I worked with hit a broken data process that cut the business off from a large block of its own customers. Couldn't reach them. Every week the process stayed broken was lost revenue—not projected, not modeled. Gone.
The honest answer from the technology side was a full re-architecture. Multiple systems, years of accumulated rules, no documentation anyone would vouch for. Months of development to do it right.
So the meeting did what those meetings do. It started deciding which timeline should win.
The Slower Timeline Wins
We've all been in this room. Technology says months. The business says now. Both statements are true, and the machinery of the company does the only thing it knows how to do with 2 true statements that conflict: it schedules a priorities meeting and makes one of them lose.
The argument about priorities is never really about priorities. It's the sound a single timeline makes when 2 kinds of work are forced onto it.
Here's the standard resolution. The rebuild becomes the plan. The business waits. Status meetings track the development timeline because that's the only timeline on the wall. And the business's urgency doesn't go away—it converts into pressure on the people doing the rebuild. Scope gets cut to hit dates. Corners get rounded. The re-architecture that needed patience gets built in a panic.
Weld a fast timeline to a slow one and you don't get the average. The slower timeline sets the pace and the faster one sets the mood. The business waits months for what it needed in weeks, and the technology team builds under exactly the pressure that produces the next broken process.
Nobody chooses this. One plan, one backlog, one owner—it looks like discipline. The weld is the default because running 2 timelines feels like running 2 projects, and everything in how we plan, fund and report pushes toward one.
Separate Path, Separate Logic
So we didn't pick a winner. We decoupled them.
On the data platform, we built a streamlined path that let the business act immediately—separate route, separate logic, visibly labeled as interim. Not a hack buried in a stored procedure. A named, documented, deliberately temporary bridge that everyone could see, including the people who'd eventually retire it.
The business was reaching customers again in weeks. And the technology team got what a real re-architecture actually requires: room. Time to untangle years of rules without a business tapping its foot in every standup. Both timelines ran at the speed their work demanded.
The label did more work than I expected. A workaround nobody can see becomes the system—we've all inherited one of those, the band-aid nobody remembers is there. A workaround everyone can see becomes a forcing function. It sat in the architecture diagram marked interim, which kept the rebuild honest: visible debt, on a schedule, with an owner.
Loose coupling was a technology idea first. Decouple your components so one failure doesn't cascade into everything else. But the most valuable decoupling on that engagement wasn't between systems. It was between the business timeline and the technology timeline. Business need and technology capacity almost always run at different speeds. The answer isn't to pick one. It's to stop making them share a clock.
The airlines that move fastest aren't the ones with the best technology. They're the ones that stopped forcing everything through a single timeline.
What this means Monday morning
If you're building the data: You're probably maintaining one of these already—an interim path someone built under pressure that never got its label. Label it this week. In the code, in the docs, in the diagram: interim, with an owner and a retirement condition. A workaround you can see is a bridge. One you can't see is just the system now.
If you're leading a data team: Go through the backlog and mark each item with the clock it's actually on—business-now or build-right. If both kinds sit in one list, the slow work is setting the pace for the fast work. Split them. Different paths, different definitions of done, and nobody has to lose a priorities argument to get moving.
If you're the executive funding the next move: Next time a fix comes back scoped in quarters, don't push to compress it. Ask what the business does in the meantime, and fund that answer as its own line. And hold both ends—an interim path without a funded rebuild isn't decoupling, it's a decision to live on the workaround. The label only holds if the real fix keeps its budget.
In my feed
A Skift test published July 6 wired Booking.com, Expedia and Viator into ChatGPT and Claude and found the connection isn't the channel. ChatGPT skipped or flatly denied apps that were live and authenticated—inventing technical excuses until the tester typed "you are wrong"—while Claude used the same connectors without a fight, so the pipes work and the disposition doesn't. For 20 years distribution meant fighting for position. In agentic channels you can be connected and still invisible, and whether your content gets used is decided inside a model nobody can audit—airlines spent a decade of NDC solving "be in the channel," and no contract yet covers "be chosen by it."
Hilton opened a direct connect with Navan on July 7, bypassing the GDS layer in corporate hotel distribution. The headline is the pipe. The story is the data—when supplier and TMC connect directly, the shopping and booking exhaust changes hands, and whoever holds the corporate booking record prices better next year. Airlines already ran this play with NDC. Watch where the record lands.
Two platform giants, 10 days, the same move. HubSpot updated its terms July 1 to share customer enrichment data across accounts—opt-out by default—and by July 5 its founder was posting "we made a mistake." Meta shipped an Instagram AI feature July 7 that opted in every public account's photos, and pulled it by July 10. Don't read the reversals as the system working. Read the defaults as the strategy. Your data inside someone else's platform is one terms-of-service email away from being their asset. That email is a commercial document now. Read it like one.
What I'm cooking
We're on Cape Cod this week. Before we left, a friend handed us a stack of flour tortillas from West Texas, and there was no world where those stayed home. So they made the trip. Fajitas outside on the grill, made queso from memory—no recipe, just taste and adjust until it matched the one in my head. The family's verdict: best ever.
The carry-on
This week I planned a rollout where the deliverable wasn't the solution. It was the sequence of rooms. Walk the hypothesis through small meetings first, a few stakeholders at a time, where people can still change their minds in front of each other. By the time it reaches the big room, it isn't a pitch. It's already the plan.
Small rooms lean toward solutions. Big rooms lean toward risk.
—Ben