Did the new routing cut delivery times?

Swiftdune Couriers (a fictional parcel company) switched its 60 courier routes to a new route-optimisation system. The contract renewal is due; as a scenario assumption, take the vendor's licence to cost about SAR 40 per route per day (an invented figure for this exercise). The operations manager asks: did the new system cut delivery times, and by how much?

Two datasets are available:

• delivery_process: for each of the 60 routes, the average delivery time per parcel over the four weeks before the switch and the four weeks after. The same routes are measured twice.
• delivery_independent: a separate random sample of 80 individual old-system deliveries and 76 new-system deliveries, from different days and routes.

The two files can give very different-looking answers. Your job is to choose the right analysis, quantify the saving honestly, and say what the study design does and does not prove. Both datasets are synthetic (generated for teaching; no real company).

Stakeholders. The operations manager (renewal decision), the finance team (licence cost), and route supervisors (who know what else changed in those eight weeks).