The first 2 months after launch
Every cloud has a silver lining.
Express scaling
In June, we started to scale the project step by step. We planned to use the service in a few test runs of our updated courses, and we also used AI Studio as the main platform for a pilot of our parallel learning format.
However, Microsoft changed its license payment policy, and we found out about it too late. Because of this, we had to temporarily stop using Minecraft Education on one of our courses. This meant that about 2,000 students lost access to their main learning platform for a few weeks.
A few weeks is a whole study module. We had to create new lessons that were still connected to Minecraft and interesting for the kids. They also needed to fit well with the academic goals of their course.
I improved our service and did some light refactoring to make it more stable. I ran load tests and used the service for the lessons that filled this gap in the program.
In the end, our scaling speed was much faster than we expected.
Metrics May-June
DAU: 1.2 → 105.1
Published projects: 23 → 1,215
Reactions per published project: 0.2 → 0.3
Share of users who open the reward box: 6.30% → 9.60%
New features and student projects
Plans for the next quarter
After such fast scaling, I had to reorganize my work. Instead of a creative mode, where I follow feedback from students and methodologists day by day, drawing a target wherever I land, I needed focused product changes. These changes should strengthen our best solutions and form an educational loop, which I described like this:
Create > Share > Iterate
If we talk about metrics instead of slogans, I need to:
Create and share — increase the number of published projects per student.
Iterate — increase the share of modified projects, meaning projects that got an update.
For a leading metric, I chose the number of reactions to a project for now. This shows how the author gets motivating recognition from other students.
Also, all new features and improvements must match new product launches. It is time for standard product frameworks. This will slow down releases and make changes, but updates once a week are still very fast compared to the months of development in the pre-AI era.