The deployment pipeline that doesn't pay back
Some teams have deployment pipelines that are masterpieces of engineering, used five times a day, and pay back daily. Others have ten times the complexity and ship once a week.
Deployment pipelines have a way of accumulating complexity that doesn't pay back. A new staging environment here, a canary stage there, a synthetic monitoring check that takes 10 minutes — each made sense at the time. Together, they make deployments take an hour to ship a one-line change.
The honest question
What does the pipeline catch that wouldn't have been caught without it? For each stage, can you name an incident it prevented? If the answer for a stage is "none that I remember," that stage is theater and should be removed.
When complexity pays back
When you ship many times a day, when the blast radius is large, when downtime costs are high. In those contexts, the pipeline is worth the cost. When you ship once a week to a system with low blast radius, you're paying for capacity you don't need.
Every pipeline stage should justify its time. If it can't, it's theater.