Garment factory line managers beat its IT system by trading workers on a handshake

The factory ran about 20 sewing lines of roughly 60 workers each on tight daily output targets. Absenteeism averaged 11 percent a day company-wide, but it landed unevenly: any single line had a 20-percent-or-worse absence morning roughly once every 11 days, and workers' home-village festival calendars meant the timing was effectively unpredictable from a planner's chair. A line that opened the morning six or eight workers short simply could not hit its quota that day, and there was no slack built into staffing to absorb it.

The factory's leadership treated this as an information and allocation problem: find out which lines are short today and move spare workers to cover them. That is a scheduling exercise, and the obvious way to run one at scale is a centralized system that takes the morning's attendance data and computes where to send whom.

The CEO hired a consultant to build exactly that: a centralized system that would ingest each morning's attendance and output an optimal worker-reassignment plan. It failed in practice because running the numbers took most of a half-day shift, by which point the shortage it was solving for was already over. Meanwhile, line managers had already been handling the same problem themselves, informally: each morning a manager short on workers would ask a small set of neighboring managers to lend one, on the understood terms that the favor would be returned the next time the borrower had spare hands.

Line managers traded workers on a handshake, lending and borrowing across lines, because the informal network compressed the response loop from half a shift to minutes.

Managers already knew who was short by walking the floor, so no data pipeline was needed.

Reciprocal favors compressed the sense-and-respond loop from half a shift to minutes.

Trust between repeat partners substituted for the oversight a central system would need.

Each transfer was small enough that lenders could absorb the cost, knowing the debt would be repaid.

When a shock hits faster than head office can respond, look for the informal peer-to-peer fix frontline supervisors already built, and widen it instead of replacing it with a slower central system.

Researchers tracking actual worker transfers found the informal peer-lending network handled shocks the centralized system couldn't reach in time, but that managers used it narrowly, trading mainly with a handful of established partners rather than the wider set of colleagues who could plausibly help. 3 percent productivity gain company-wide from absence shocks alone -- a modest-sounding number that, compounded across every line every day, is the gap between the informal fix as it existed and the informal fix at full strength.