Some of the best insights about how labs work come from people who are trainees with me. I do my best to encourage a "flat lab", with as little hierarchy as possible. Someone (carpet people, from the part of the university where there are carpets on the floor, as opposed to us linoleum/tile people) recently was shocked that I encourage the medical students to call me by first name in my lab. What the heck. That is a very little thing. There are much bigger things: like actually listening to these students in lab meeting, when planning and treating their ideas and thoughts with respect or criticizing a particular idea or strategy and not the person who said it. My thesis advisor once told me (something like): I can't believe you brought this piece of trash to me and are wasting my time. That was over 35 years ago, and I still remember it. But he's dead.
One of the insights, directly from trainees that I have seen over the years is how uncomfortable "business operation" labs make them. If they wanted to train with a postdoc instead of a PI they would pick the postdoc, not the particular lab. This is likely a reflection of the people who are attracted to my particular brand of lab-chaos or people who care less about being in a shiny lab or doing "cutting edge science" and more about learning something.
I am a fan of "meta-rules", over-arching guidelines that pre-empt having lots of little rules to think about. For grant writing, the meta-criterion is "make the reviewer your advocate", or more aptly "do not piss off the reviewer".
For running a lab and mentoring, one meta-criterion is treat your people like human beings, the way you would want to be treated. You'll make mistakes, we all do and have and will continue to do. But seeing people as people will go a long way to making things work.