Draft, open for argument

Find the constraint first

Work piles up in front of exactly one station. Improve that one.

The rule

A delivery line ships as fast as its slowest station. Spending on any other station changes nothing you can measure, and it feels like progress, which is worse.

Signs you are breaking it

  • A large budget for coding agents while pull requests wait two days for review.
  • Dashboards that show activity per engineer and nothing about queues.
  • A tooling roadmap ordered by what is exciting.

In Lights Out

Buy an optimization away from the constraint and the log tells you ten days later that the needle did not move. The money shows up as early spend.

Back to all factors