A 140-person offsite, two airports, one thunderstorm and a plan already drafted

The situation

A typical event: 140 people, three days at a mountain lodge, arrivals spread across nine hours at two airports. The shuttle plan lived in a spreadsheet that took most of a day to build. It was right until the first flight slipped.

From there, everything happened in a group chat. A morning delay reshuffled three vans. An afternoon thunderstorm closed the ropes course and the river float, and 60 people needed somewhere to be in twenty minutes. The lead planner usually found an answer, because lead planners are good at this. But the answer lived in one person’s head and one person’s phone, and every event started from scratch.

What we tried first

We first optimized shuttles for seat utilization alone. On paper it was beautiful: two fewer vans per event. Replayed against a past event, it would have left the day’s first arrivals waiting 70 minutes for a full van. Utilization is a cost to manage, not the goal. We made maximum wait a hard constraint and let utilization settle where it could.

We also tried fitting the firm’s disruptions into a general-purpose incident tool. It tracked incidents well and helped resolve none of them. The firm didn’t need a ticket. It needed the indoor alternative, the new van times and the attendee message ready before the storm arrived.

What we built

Group routing with people in it. Route Engine batches arrivals into shuttles within a 40-minute maximum wait, holds accessible seats for attendees who need them, and accounts for walking times inside the venue, which matter more than you’d think at a lodge spread across a hillside. Activity assignment respects capacity, stated preferences and energy budgets.

Playbooks prepared before anyone travels. For each event, the firm and our team pre-approve alternatives for nine scenarios: which indoor sessions can absorb the ropes-course group, which restaurant can take an early dinner, which vans can be added at what notice. Triggers come from flight status, weather alerts and supplier messages.

Drafts, approved in one step. When a trigger fires, Route Engine re-plans the affected part of the day. The event lead sees the revised schedule, the van changes and the attendee message together, and approves or edits from a phone. Nothing reaches attendees without that approval.

A review after every disruption. Within 48 hours, the team walks through what triggered, what was proposed and what was changed. Playbooks get better between events, not only during them.

What changed

Shuttle plans went from a day’s work to about an hour of review. Vans ran fuller and fewer people waited long at the curb, both at once, which the spreadsheet never managed.

The bigger change came in the afternoons. During the third live event, a storm cell built over the ridge at 2:10 PM. The indoor plan, revised van times and attendee message were drafted by 2:14. The event lead approved at 2:19. The ropes-course group was inside before the first thunder.

What we’d do differently

Drill before the first live event, not after. We ran the first tabletop drill after event one. It surfaced two gaps in the approval path that we were lucky not to need.

Measure venue walking times on site. We estimated the walk between lodge buildings at five minutes from a map. It was twelve, uphill. One planner with a stopwatch on day one would have fixed three schedules.

Make the manual override obvious. Planners sometimes needed to pin a van or an attendee by hand. The override existed but took three taps to find. It now takes one.