👉 Always respect the progressive rollout, but why?

📆 A long time ago, I heard that something awful happened because a company didn't respect the progressive Rollout; in other words, a friend told me that he was launching a new feature, and the team had a very detailed plan that was doing a long progressive rollout to understand the users' behavior.

The plan was something like this, launch the feature and:

➡ in the first two weeks to 2% of the user base;

📊 Week 3 ↔ Keep the 2% and separate one week to study the users' behavior against the cohort;

➡ Week 4 ↔ Rollout the feature to 5% of the user base;

📊 Week 5 ↔ Keep the 5% and separate one week to study the users' behavior against the cohort; the new feature is doing a great job, so let's go on;

➡ Week 6 ↔ Rollout the feature to 15% of the base;

📊 Week 7 ↔ Keep the 15% and separate one week to study the users' behavior against the cohort; the new feature is doing a great job, so let's keep this pace;

➡ Week 8 ↔ Rollout the feature to 30% of the base; 📊 Week 9 ↔ Keep the 30% and separate one week to study the users' behavior against the cohort; the new feature is doing a great job, so let's proceed with the good work; ❌ Week 10 ↔ A high-level person in the company decided that it was enough and ordered to release of the new feature to 100% of the base, which was ultimately against the plan.

And then, what happened?

❌ Problem 1 ↔ the company didn't have enough infra to support all data that was coming, so all the company application was dropped down and stayed off for one week; ❌ Problem 2 ↔ 1 week off, a lot of money was lost, partners and users got angry; ❌ Problem 2 ↔ Most of the 70% of the users base didn't like the new feature, so the application rate went down due to the new feature and because it stayed off for one week; ❌ Problem 4 ↔ When the new feature was released to the base, the data team discovered another behavior, 70% of users they prefer to use other applications to do the same work that the new feature was doing; ❌ Problem 5 ↔ The company had to create a task force with many developers to solve the problem, so they stopped to work on what they were doing previously, which affected the roadmap of several teams.

🔆Conclusion: Of course, you can have some luck and maybe roll out a new feature to 100% of the base would work, but know that most of the time, it don't, so prevention is better than cure; in other words, you need to learn from your mistakes and always try to respect the progressive Rollout, mainly if the company has a vast user base, the behavior of 30% of your users is not the same as 70%.