Hi Railway team,

I maintain several templates in the marketplace and I'm trying to understand how the template health score is calculated.

My MySQL template dropped from 75% to 60% and then to 50%, and the last drop happened without a single new deployment. The total deploy count stayed at 46, the template config hasn't changed since Oct 28, and a fresh test deploy today ran fine, including a redeploy on an existing volume. It looks like older successful deploys roll out of the time window while a single failed one stays in. With such a small sample, that swings the score heavily.

A few questions and suggestions:

1. What exactly goes into health? What is the time window, are deploys from the template author's own workspace counted, and what counts as a failure (build, deploy, crash after start)?
2. Could you smooth the score for small samples, for example with a minimum number of deploys before a score is shown, or with weighting?
3. Could you show the number of deploys behind the score, and ideally the reasons for failed deploys? Then authors could actually fix whatever is breaking for users.

Right now a single failed deploy by a user, possibly caused by their own setup, can make a working template look broken, and that hurts discoverability. More transparency would help a lot.

Thanks!

André