The teams with the slowest release cycles almost never lack automation — they have pipelines, just ones that have quietly accumulated years of unaddressed friction that nobody has the mandate to fix.
Flaky tests are the most corrosive pattern. Once a team learns to re-run a failed pipeline instead of investigating it, the entire test suite's signal value collapses, and real regressions start slipping through with everything else.
Pipeline duration matters more than teams give it credit for. A 45-minute pipeline doesn't just cost 45 minutes — it changes developer behavior, encouraging batched, riskier commits instead of the small, frequent changes that make CI/CD valuable in the first place.
Approval gates deserve regular audits too. Change advisory processes that made sense for a quarterly release cadence often persist unchanged into a daily-deploy world, adding friction without adding safety.