# Nobody shops a new recipe at Thanksgiving. Why do it in prod? #devtools #developertips #devtips

## Метаданные

- **Канал:** LaunchDarkly
- **YouTube:** https://www.youtube.com/watch?v=dvhrnQ1S7Xc
- **Дата:** 08.04.2026
- **Длительность:** 1:17
- **Просмотры:** 880

## Описание

Alexis breaks down progressive rollouts using the best analogy we have ever heard. Ship to 1% first, then 5%, then 25%, watching for error spikes at every step. Add a kill switch as a safety net and user targeting for precise control, and you have a system that lets you ship fast without losing control.

Want a step by step tutorial? Check out the full blog here → https://bit.ly/4cfOZVB

#progressiverollout #featureflags #featuremanagement #devops #softwaredevelopment #softwareengineering #cicd #continuousdelivery #releasemanagement #LaunchDarkly #killswitch #usertargeting #cloudcomputing #productengineering #techshorts

## Содержание

### [0:00](https://www.youtube.com/watch?v=dvhrnQ1S7Xc) Segment 1 (00:00 - 01:00)

You ever cook a recipe for the first time and immediately serve it for Thanksgiving or some big family event? I ain't think so. See, know you have to taste it first and get some other people to try it and then maybe you can bring it to a family function. That's a progressive roll out. Instead of shipping a new feature to every single user at once and hoping for the best, you release it to a small group first, maybe 1%, then five, then 25. And at each step, you're watching. Watching for what, you may ask? if there's any spikes in error rates. You want to know, is the user experience actually good? Let's say you rolled it out to 10% of your users, the same set of users every single time. This makes it a more consistent, controlled experiment. Now, pair that with a kill switch, which is a feature flag that is designed to shut down everything to turn the faucet off so that you can go and fix the root cause. It's like having a safety net under your safety net. And let's say you want to get even more precise. You can use what's called user targeting. That's where you can control who sees the feature based on things like account type, user ID, location, or whether they're on your internal team. You've got a system, small steps, real data, while still maintaining full control. Because shipping fast is cool, but shipping fast and remaining in control, that's even

---
*Источник: https://ekstraktznaniy.ru/video/47484*