Logistics optimisation

Context

International parcel shipping company

Problem

15% to 20% of address data needed manual identification which required a lot of resources. After manual input, 4% of the packages were still delivered to wrong address. The geographical location of delivery points was frequently inaccurate and the delivery windows very broad, leading to undelivered packages.

Solution

The availability of a lot of data and IoT handheld devices at point of delivery gave us the means to make parcel delivery more efficient, while transferring knowledge to the clients IT team. We designed and built models for each task while getting familiar with the specifics of the domain. In close collaboration with the in-house IT team we were able to move from experiments to production deployed solutions in fast cycles.

Results

  • Reduced manual intervention of address data by 90%
  • 50% reduced delivery failures
  • 2000 direct man hours saved per month (less manual correction)
  • 5000+ indirect man hours saved per month (post error correction handling)