Esmail Arshad
All case studies

Amazon · Supply Chain Program Manager · 2021 — 2022

Tracing a hidden cost anomaly on a Southeast freight lane

Daily monitoring exposed a cost pattern that weekly reporting had hidden, leading to a fraud investigation and preventing further losses.

Tableau · Advanced Excel · Internal TMS

Signal
~10% of Southeast shipments running at roughly 3× expected cost
Investigation
3 months of shipment-level data reviewed
Exposure
~$360K three-month financial exposure identified
  1. Southeast region shipments

    ~100 shipments per week

  2. High-cost shipments

    running at roughly 3× expected cost

  3. Atlanta–Miami lane

    where the cost spikes concentrated

  4. One repeated facility pair

    same origin and destination on each affected shipment

  5. Same receiver and driver

    pattern confirmed across three months

A regional cost spike narrowed to one lane, one facility pair, and the same individuals.

Challenge

Ahead of Amazon’s newly launched October Prime Day event, Amazon Freight performance monitoring moved from weekly to daily. The weekly cost and delivery KPIs still looked healthy. The daily view did not: cost per mile in the Southeast was spiking, and roughly 10% of shipments in the region were running at about three times their expected cost.

What I did

  1. Compared on-time against late shipments to establish whether the cost variance tracked service failures or something else.
  2. Broke the data down by lane, mileage and sortation facility in Tableau and Excel, which narrowed the anomaly to the Atlanta–Miami lane.
  3. Reviewed three months of shipment IDs, pricing rules and TMS notes, moving from regional performance down to individual transactions.
  4. Identified the same facility pair, receiver and driver recurring across the affected shipments.
  5. Compiled the evidence and escalated it through management for investigation by the relevant teams.

Decision

Weekly performance was still inside target. I treated the daily cost spikes as a real signal rather than noise, and kept narrowing the data until the pattern was explainable.

Outcome

The investigation confirmed fraudulent activity.

Each affected shipment carried roughly $3K of excess cost, amounting to about $360K of financial exposure over three months. The activity was stopped and further losses were prevented.

All case studies