Performance bot
Mendix enables app building with low-code, making it easier and faster to build what users need. App users have little patience for slow load times and connection issues. Application performance is a key factor that impacts a pleasant user experience. Performance Bot offers suggestions for improving app performance, based on machine learning and anonymous usage data.
Collaboration
What made this project unique was the collaboration that happened. My aim and role was to guide them towards and through a truly collaborative process, with PM, Devs, Design and UXR. People were hesitant at first to do all activities as a team, doubting the added value and worried about time investments. But I insisted and it paid off.
Starting from OKR's
We started from an OKR goal to increase adoption of Performance Bot with 10%. Based on this goal we defined hypotheses, and then planned qualitative research to test them. We investigated with 5 users who were using the Bot, as well as with 5 who weren't, and spent one-hour with each of them.
Valuable outcomes
The shared sense-making of all our insights brought clarity on the issues with Performance Bot:
-
lack of awareness and lack of trust
-
confusing Logic Bot with Performance Bot
-
not aware that anti-patterns and best coding practices were powering the Bot.
Product direction
Fueled by our insights we brainstormed on problem oriented capabilities:
-
adding a clear place for 'Best Practices' in a new tab
-
renaming Mendix Assist (which later became Maia - Mendix AI Assistance)
-
improved explanation about the AI -powered performance best practices, both on the Mendix website and in the documentation.
The collaborative approach lead to fewer roadmap and planning discussions afterwards, because everyone had been involved from the start.
But this project also brought other gains: It increased team cohesion, gave the team more focus, and increased accountability.