Margin up 15 points, with better quality

The problem

An experienced manager checked every picked cart, item by item. QC took about 13 minutes per order, and roughly 7 in 10 carts had at least one error. At the planned 80 to 100 orders a day, QC would need seven full-time checkers per warehouse and about 1.6 labor-hours per order. Even at low volume, a full aisle of carts sat waiting for QC.

The root cause

QC was applied to every cart and every picker because there was no data on who was reliable. Without picker-level accuracy history, the only safe policy was to check everyone, every time. Quality was inspected at the end of the line instead of built into the pick, so its cost grew with every order.

What we did

The app started tracking each picker's error rate and speed, and a bonus paid on both, earned only when a cart shipped clean. The fastest way to earn more was to pick accurately.

Once a picker's error rate held under 1%, the app moved them to "super-picker" status and checked 1 in 10 of their carts. If a sample slipped past about 1.5%, they went back to full QC until they recovered. New pickers stayed at 100%. We also changed the app to pick by bin location instead of category, and automated the paperwork for out-of-stocks and cart labels.