A 3-Year Record of Building a Code Review Culture
When I joined the team last year, code review was practically a formality where people just left 'LGTM.' I spent three years trying all sorts of things to change that, and now it has more or less taken root. I'm writing this up mainly around the trial and error.
1. I didn't start with rules
At first, I wrote a review guideline document, but nobody read it. So instead of a document, I switched to sharing good examples from actual review comments.
2. I didn't assign reviewers
There was a problem of work piling up on specific people, so I had any two people other than the author take the review. Surprisingly, this worked well.
3. I standardized the comment tone
When I changed 'Why did you do it this way?' to 'I think doing it this way could cause a problem—what do you think?', defensive replies dropped dramatically.
4. I treated review time as work
This was the biggest one. Once review time was recognized as official work hours, everyone started taking it seriously.
Looking back, I think it was 80% cultural rather than technical. I hope this helps anyone dealing with similar concerns.
6 answers
Agreed, especially #4. If review time isn't recognized as work, you end up doing it after hours, and then you just skim it. This one thing seems to make or break the culture.