01/2026.03.14/Engineering/6 min read

Stop staging, start shipping

We deleted our staging environments. Production is fine. We have receipts.

We don't have a staging environment. We deploy straight to production, multiple times a day, on every project we run. This is going to make some of you angry.

Staging exists because someone got burned in production once and decided what they needed was a copy of production where they could be wrong privately. The problem is the copy is never actually a copy. The data is fake. The traffic is fake. The integrations are stubbed. So staging tells you everything is fine, and then production tells you that staging was lying.

We replaced staging with three things. Feature flags so half-finished work can ship dark. Observability so we know within sixty seconds when something is broken. And a rollback that takes one command. None of that is novel. Big tech has done it for fifteen years. The novel part is doing it as an eight-person studio on small budgets.

The first time we shipped to prod without staging was scary. The second time was less scary. By the tenth time we'd shipped twice a day for two weeks and the world had not ended.

Here's what nobody admits about staging: it's also where work goes to die. A feature gets to staging and then it sits there. Someone is going to test it. Someone is going to QA it. Someone is going to schedule the prod deploy. By the time it ships the engineer has moved on to something else and isn't holding the ball. The longer the runway between “done” and “shipped”, the more our brains turn done into a finish line. It isn't.

If you're building software and you're not shipping daily, the bottleneck is probably not your engineering team. It's the staging environment, the QA queue, the release manager, or the person who needs to give it a final look. We removed all three. We have not lost a client. We have not had a meaningful prod incident in fourteen months. If we do, we'll fix it in production, like adults.

If your team can't ship to prod safely, the right move is not a bigger staging environment. It's better tests, better flags, better observability, and a culture that treats deploys as boring. Ours is so boring our engineers usually deploy from a coffee shop.

Anyway. Tear down your staging. Or don't. We're not your boss.

Brigada.dev
Prishtinë, Kosovo
← Back to home
More from the journal
02A grid is not an opinionDesign03On-call without tearsOps04The SEO work we stopped doingGrowth