A warehouse automated a third of its picking. Headcount fell by 4%.

A three-year study of eleven distribution centers that installed goods-to-person automation found the effect on employment was much smaller than either advocates or critics predicted, and the effect on the composition of the workforce was much larger.

Across the sites, 31 percent of picking tasks were automated. Total headcount fell 4 percent. The gap between those two numbers is the finding.

Where the people went

Automation removed the routine walking-and-picking portion of the work almost entirely. It did not remove the exceptions: damaged cartons, mis-slotted stock, items the system cannot grip, orders that must be broken and rebuilt. Roles handling exceptions and maintaining the equipment grew 39 percent.

Those jobs pay more. They also require training that the previous roles did not, and the study found the sites that transitioned existing staff into them retained workers far better than the sites that hired externally.

We did not need fewer people. We needed different people, and it turned out the best ones were already on the floor.— An operations director at one of the sites studied

The hiring problem

The constraint that surprised operators was maintenance technicians. Vacancy duration for those roles averaged 94 days across the sites, against 11 days for general warehouse positions. Two sites ran below their target staffing for the entire study period.

• Sites studied: eleven distribution centers over three years
• Picking tasks automated: 31% on average
• Headcount: down 4%; exception-handling and maintenance roles up 39%
• Maintenance technician vacancies: 94 days average, versus 11 for general roles

The authors are careful about generalizing. All eleven sites were growing throughput over the period, which means automation absorbed growth rather than displacing existing volume. In a flat or shrinking network, they write, the same technology would likely produce a considerably larger headcount reduction.