At the retro someone said it out loud. We never tell each other what is working or what is not, we just wait for reviews. Everyone agreed, and your lead turned to you. Could you set something up?
It is a good job to be given, and a surprisingly easy one to get wrong. Most peer feedback schemes fail in one of two ways. They become a form nobody fills in, or they become a quiet way of collecting complaints for performance reviews. Your design should avoid both.
Keep it separate from ratings. Say clearly, and get your lead to say it too, that what people write to each other goes to each other and not into anyone's review. The moment people suspect it feeds a score, they start writing politely and carefully, which makes it useless.
Make it small and regular rather than large and rare. A short exchange every six weeks is far better than a long survey once a year. Pair people up, rotating each round, so everyone gives and receives with someone they have actually worked with recently.
Give people questions that produce something usable. What is one thing I did recently that helped you? What is one thing I could do differently that would make working with me easier? Two questions, one concrete example each. Open boxes labelled strengths and weaknesses produce adjectives, and adjectives cannot be acted on.
Suggest a conversation, not just a written note. Fifteen minutes on a call, exchanging both answers, with the writing as preparation. Spoken feedback between peers is easier to soften and clarify, and easier to thank someone for.
Go first, visibly. Ask a colleague for feedback in the team channel and share, briefly, what you are going to change. People copy what they see more readily than what they are told.
Make it optional at the start, and check after two rounds. Ask the team whether it is helping and what to change. A scheme that the team adjusts is one they will keep.
And expect the first round to be cautious. Trust grows round by round. That is the scheme working, not failing.
– Asael Shinder
