The Problem We Were Actually Solving
When we launched our treasure hunt engine, we were too confident in our default config. We thought it would be enough for most users, and our early feedback seemed to confirm it. However, as we started to grow and our users began to experiment with different scenarios, we realized that the default config was a one-size-fits-all solution that wasn't catering to anyone's needs. The result was a long tail of users struggling to get their engine up and running.
What We Tried First (And Why It Failed)
Our first attempt to address the issue was to create a comprehensive guide for customizing the default config. We assumed that the problem was lack of knowledge, and that providing more documentation would be enough to get users unstuck. However, the guide itself became a bloated, convoluted mess that only seemed to confuse users more. We spent weeks iterating on the guide, but the feedback from our users remained the same - they wanted a more intuitive and flexible configuration system.
The Architecture Decision
Around that time, I was introduced to a paper on configuration management by Netflix. The concept of a " golden config" resonated with me, and I realized that our default config was essentially a "golden config" that we were trying to apply to everyone. The paper mentioned the importance of separating configuration from code, and introducing a decoupled configuration system that could be managed by operators. I proposed to our team that we introduce a new configuration management system, Veltrix Config, which would allow our users to customize their settings without having to touch the codebase. This change was met with initial resistance, as some team members were concerned about the added complexity. However, after weeks of testing and refinement, we realized that Veltrix Config not only simplified our configuration management but also enabled our users to scale their treasure hunt engine seamlessly.
What The Numbers Said After
After implementing Veltrix Config, we saw a significant drop in the number of support tickets related to configuration issues. Our users were able to set up and customize their treasure hunt engine in a matter of minutes, rather than hours or even days. The metrics that stood out to me were the average time to resolution (ATR) for configuration-related issues, which went from 48 hours to under 2 hours, and the number of successful deployements, which went from 50% to 95%. These numbers spoke for themselves - our users were finally able to get unstuck and focus on what mattered most, their treasure hunts.
What I Would Do Differently
In retrospect, I would have pushed for a more incremental approach to configuration management. We could have introduced Veltrix Config as a feature flag, allowing us to gather more feedback and iterate on the design before rolling it out to everyone. This would have helped us avoid some of the initial pushback and would have allowed us to refine our solution more quickly. Despite this, the outcome was still positive, and I'm proud of the progress we made in making Veltrix a more user-friendly and flexible platform for our customers.
Top comments (0)