This applies to all aspects of life and work. Figuring out the mix of direct and behavioral feedback that lets you learn about your interactions with others (and predict/manipulate future ones) is perhaps the single task that has selected for human thinking potential over the millennia.
One key is looking for objective behavioral data, and using that to guide your discussions/interviews about what the feedback-givers tell you. How many times have you listened to it? Did the lyrics get repetitive, or was there enough depth to keep you interested? Do you listen to the end, or play just a segment? What takes you out of the moment when you listen? Etc.
I got this idea from our music-makers chat.
Imagine you record a song and your friends say it's "nice". What does that mean? Is it really nice? One way to solve the problem is to see what feedback your friends give to songs recorded by your other friends. For example, Bob shares some violin playing that is obviously (to you) off key. Carol says "it's nice, but the intonation on some notes could be a bit improved". By integrating a bunch of evidence like this, you can learn two things: 1) how to correctly interpret feedback given to you, instead of taking words like "nice" at face value 2) how to give feedback to other people, without seeming either too blunt or too diplomatic.
The same idea applies to many things besides music. We can call it "feedback calibration". I'm pretty sure it isn't new, many people do it intuitively anyway. But maybe seeing it spelled out will be useful to someone.