The thirteen hours nobody budgeted

Stripe's Developer Coefficient study surveyed thousands of engineers and executives and produced a number that still organizes budget conversations: developers reported spending about 17.3 hours of their week on maintenance work. The detail that matters more than the headline is the composition: 13.5 of those hours were attributed to technical debt specifically, which means debt is not a slice of the maintenance burden. It is most of it.

The study is from 2018, which is worth saying out loud, and we know of no reason the shape has improved since; every diagnostic we have run finds the same silhouette, whatever the exact hours. Roughly a third of engineering capacity servicing the past, most of that serving decisions everyone regrets and nobody scheduled fixing.

Two things follow if you take the number seriously. First, hiring does not buy what leaders think it buys: add ten engineers to a system with this profile and you have hired roughly seven, before onboarding. The capacity plan should say so. Second, debt paydown is not a virtue project competing with roadmap. It is capacity recovery, and it can be costed like any investment: hours spent against recurring hours returned, with the estimate written down and checked later.

The tradeoff of taking it seriously is a roadmap that visibly slows this quarter to speed the ones after, which is a hard sentence in a planning meeting. The condition that reverses the advice is a young codebase that genuinely does not have the profile yet. Check first: your ticket history knows which one you are, even if the roadmap deck does not.