If you were watching this board on Saturday, July 18, you saw it happen in real time: the delay bars going orange, then red, the cancellation count climbing, Dulles and Newark seizing up. Here's what was behind it. At about 6:00 a.m. Central, SHARES — the passenger service system that underpins United's check-in, boarding, and reservations — went down. It was restored by roughly 7:15 a.m. Seventy-five minutes of downtime. The disruption it caused lasted the rest of the weekend.
The mechanics of why are worth understanding, because they're the same every time. When the reservation system is down, airplanes are fine and crews are fine — but nobody can be checked in, bags can't be tagged, and boarding stops. Every departure in that window leaves late or not at all. Late aircraft miss their next turn, crews time out, and the schedule starts eating itself. Do that at 6 a.m. — the morning bank, when the whole day's schedule is stacked up and ready to launch — on a summer Saturday, and you've poisoned the entire operation before breakfast.
Then the weather arrived. Severe East Coast storms hammered exactly the hubs that were already wounded, and the two failures compounded. By day's end the U.S. system had logged about 2,437 cancellations — to be precise, that's the industry-wide, all-causes number for the day, not a United-only figure. United's own share was roughly 268 cancellations, about 8% of its schedule, plus the worst delay rate of the five biggest carriers: about 39% of its flights ran late. The hardest-hit stations were Washington-Dulles, Newark, and San Francisco — two of them the same hubs United has spent the year holding up as operational success stories.
The uncomfortable part of this story is the system at the center of it. SHARES is old — a Continental inheritance that United has been patching and extending for well over a decade, and the airline has had a string of high-profile outages on it over the years. Every airline has IT failures; Delta's 2024 CrowdStrike recovery took nearly a week. But there's a reason "SHARES outage" is a recognizable phrase to United frequent flyers in a way that has no real equivalent at some competitors. An airline spending billions on Starlink, seatback screens, and new interiors is still running its most operationally critical software on foundations from another era.
The honest framing: United's recovery, by meltdown standards, was actually decent — one bad Saturday and a rough Sunday, not a week-long cascade. But for an airline whose entire premium strategy depends on being the reliable choice, the lesson of July 18 is blunt. The product in the cabin has been transformed. The plumbing underneath it hasn't — and on the wrong Saturday, the plumbing is the product.