Sixteen Months of Unpaid On-Call, Then Two Resignations
Three engineers carried unpaid on-call for sixteen months. Two resigned the same quarter; the third asked off. A stipend would have cost far less.
Three engineers ran our production on-call for sixteen months. Weekly rotation: one primary, two backups. We paid them $0 for it. The comp philosophy filed on-call under senior engineering, and the three of them accepted that at hire. Around month nine, they stopped accepting it. They never said so out loud.
Two of them resigned the same quarter. The third asked off the rotation. Each gave a different reason on the way out. In private, the real one was identical. On-call had eaten weekends, holidays, and the slack time that quietly compounds when no one pays for it. Sixteen months was what they would absorb unpaid. After that, the arithmetic broke.
Unpaid on-call is one of the most reliably mishandled retention risks at a growing company. The company banks the reliability. Production stays up, incidents close, customers stay. The engineers eat the cost: broken sleep, canceled travel, the low hum of always being reachable. That cost sits on no dashboard. The trade stays invisible right up until someone prices it and leaves. It rhymes with the quarter that ran on heroics and the day a system finally replaced a hero: the org leans on goodwill until the goodwill files its notice.
Paying for it is not complicated. Put a per-week stipend on the primary slot. Per week, not per incident, because you are buying availability, not counting pages. The number moves with stage and page volume. A few hundred dollars a week of primary rotation is ordinary. The dollar figure matters less than the signal: the work costs something, and we are willing to pay it.
Most companies under a couple hundred engineers pay nothing. The reasoning is some blend of "senior people are supposed to do this" and "we don't have budget." The first is the story that manufactures the resignations. The second collapses the moment you set it next to the cost of replacing one senior engineer.
Do the math on ours. Three engineers at four hundred dollars a week, sixteen weeks of primary each, runs about nineteen thousand dollars. Replacing two of them cost us north of three hundred thousand, counting recruiting and ramp. The stipend loses to nothing here. Most companies learn this only after the exits, and most still skip the stipend, because they book the departures as personal choices instead of a bill the org declined to pay.