With the advent of high-performance black-box models, interpretability is becoming a hot topic today in machine learning. While a lot of research is done on interpretability, machine learning researchers do not have precise guidelines for setting up user-based experiments. This paper provides well-established guidelines from the human-computer interaction community.