"No. I'm saying whoever approved the design may have been measuring the wrong outcome. A system can work exactly as designed and still be wrong for the job."
Marcus's expression changed. He slid a battered notebook across the desk. Inside were handwritten delivery windows and notes about customer exceptions. Several pages were covered with calculations more accurate than the official schedules.
"I keep my own list," he said. "It saves me from making the same mistakes the computer makes. But I'm not supposed to use it because somebody upstairs decided handwritten notes are unprofessional."
Daniel leafed through the notebook. He did not see an old dispatcher refusing to adapt to technology. He saw a man quietly running a second planning system because the official one could not be trusted.
"Would you help me test a better model?" he asked.
Marcus hesitated. "Am I going to get blamed if it doesn't work?"
"We'll document the changes together. If there's a failure, we'll investigate it together."
"That's not how this place usually works."
"Then let's make it how one part of the place works."
For the next three evenings they remained after the dispatch rush to compare the planned routes with actual delivery times. Daniel had no interest in pretending to write code he did not understand, so he brought in a routing analyst and a systems engineer from the support team. Marcus provided the operational logic, Daniel secured the time and authority to test it, and the analysts built a controlled pilot. They began with a limited region rather than changing the entire network overnight.
The first attempt failed in an instructive way. It improved delivery windows but increased driver idle time because it failed to account for how some customers unloaded. Marcus spotted the problem before the end of the first day. Instead of accusing the drivers of inefficient behavior, they adjusted the model and repeated the trial.
After two weeks, late deliveries on the pilot routes were down thirty-one percent against the prior comparable period. Daniel asked Meera's finance team to verify the figures independently. They confirmed the improvement and found that fuel costs and overtime had also fallen in the trial area. Three weeks later, the gains had persisted and extended to neighboring routes as the team refined the schedule.
Daniel brought Marcus into the executive review. The dispatcher arrived wearing a clean but worn shirt and looked uneasy amid the glass walls. An analyst offered to present his findings for him. Daniel shook his head.
"Marcus identified the problem. He should explain it. I'll handle the questions."
Marcus stood at the screen and pointed to two identical route maps: the old version, clean and short, and the revised version, less elegant but better aligned with customer receiving times. He explained why a shorter line on a map did not always represent a faster or better delivery. By the end of the meeting, three executives had taken notes on his recommendations. Meera spoke first. "Why wasn't this issue identified during the software rollout eighteen months ago?"