I believe in a fast iteration based on user feedback rather than building the perfect product in isolation.

There are two sides to launching a product to production: your app performance and the user experience.

You need good visibility into both.

Your app performance:

This is your performance benchmark. It shows whether the app architecture and the infrastructure are up to standard.

We're looking for efficient operations without errors and within expected cost parameters for performance.

• CPU utilization, RAM usage, Number of containers
• Log analysis - tracking error conditions
• Load times and response times
• API requests/second

Without these metrics, you are flying blind. But even with these numbers, the app is a black box.

If you can't actually see the users playing, how do you know if they are having a good time?

You need to track signals.

The user signals:

• Conversion rates from ad clicks
• How long do they keep playing
• How often do they come back
• What actions do they trigger
• How far do they get

This is not a comprehensive list; it depends on what you want to test and what you care about.

I already had all some solid metrics and dashboards in place for the infrastructure before the first public test, but yesterday and today, I spent building out the second part.

Gaining visibility into user behavior by writing db queries, parsing Redis sessions, and looking at Google Analytics.

This is where learning comes from. Public tests won't tell you much unless you collect solid data that can back up or refute your hypothesis.

Yours,

Taj