When I was an engineering manager at Google, my group would look more closely at impact than we would at output. In fact, if the output ended up being more tech

When I was an engineering manager at Google, my group would look more closely at impact than we would at output. In fact, if the output ended up being more technical debt than durable improvement (the proverbial "I'm gonna write me a new minivan this afternoon!" effect that Dilbert spelled out), we would count that against the engineer's evaluated performance. And it would usually take more than 2 months to determine when that was the case.