We're going to do what we've been doing (profile & find issues OR pull from the backlog) and additionally, I'll populate the backlog with relevant things focusing on our goals, such as:
Choose a start up config to profile each week
Ensure we’re profiling and not just doing the backlog
Figure out how to share hunches (i.e. not known improvements but close)
Regularly examine telemetry -> file new issues based on it
Regularly re-evaluate testing needs -> file new issues based on it
Proposal: https://docs.google.com/document/d/1ydgMIhMlD4KXt09SWDO4QxHqGpMf6F58P_RvPlqB8Kk/edit
I'm trying to figure out how to put it into practice now.