Sixty-two depots on one plan, in a fortnight

The situation

Kestrel moved 4,000 pallets a day across sixty-two depots, planned in two spreadsheets that only one person fully understood. When he was on holiday, on-time performance fell by eleven points, every year, predictably.

The failure that finally forced the issue cost eleven thousand euros in one afternoon: four depots reported the same pallet and three trucks were sent to collect it. Nobody had made a mistake.

What we did

We started with the six depots that complained loudest rather than the six that were easiest, on the theory that anything surviving them would survive the rest.

• Modelled depot capacity from two years of historical loads
• Replaced the overnight spreadsheet export with a live source
• Wrote one page of training, tested on a new starter
• Left the spreadsheets running in parallel for three weeks

The parallel run mattered more than anything else. It let the planners check the new numbers against instinct until instinct stopped disagreeing.

What changed

Duplicate collections effectively stopped. On-time drops rose from 61% to 86% within two months and have held there since. Overtime fell, because the five o'clock scramble largely disappeared.

The measure Priya cares about is different: a planner in their second week now makes the call that used to require six years of experience.