The price of mandatory code reviews

The post questions mandatory code reviews and uses data from 400+ companies and 3,000+ engineers to compare speed, bugs, review quality, and turnaround time. It argues that skipping reviews increases output but also raises bugs, while selective, fast, high-quality reviews can preserve both speed and quality.

Reading notes

#

• Teams without code reviews ship faster, but they also have far more bugs.
• More reviews are linked to fewer bugs, with the biggest gains between zero and 0.5 reviews per PR.
• Higher-quality reviews slow teams down, but reduce bugs substantially.
• Review speed matters because PRs reviewed in under 3 hours are much more productive than PRs that wait 8+ hours.
• Top teams are faster and still keep bugs per feature lower than median teams.
• The quality of reviews people give tends to match the quality of reviews they receive.
• The author does not want to abandon code reviews, but plans to make them selective, high-quality, and fast.