Data-Driven Engineering Insights
Original research based on millions of GitHub pull requests. Real numbers, actionable insights.
Featured Research
2025 Engineering Benchmarks: Year in Review
What changed in how software ships—and what to watch in 2026
GitHub Archive / BigQuery | October 2025
83%
No Review
(1000+ line PRs)
68%
Self-Merged
(all PRs)
18x
Less Scrutiny
(large PRs)
53%
Longer Wait
(new contributors)
Deep Dives
3.4M
PRs Analyzed
3.4 Million PRs, One Uncomfortable Truth
Analysis of 3.4 million merged pull requests reveals 83% of large code changes ship without review. The largest public study of code review practices on GitHub.
18x
Less Scrutiny
The Rubber Stamp Problem
Large PRs receive 18x fewer review comments per line. The bigger the change, the less anyone looks.
68%
Self-Merged
The Self-Merge Reality
Two-thirds of all code on GitHub is merged by the same person who wrote it.
53%
Longer Wait
The First-Contribution Tax
New contributors wait 53% longer for their PRs to be reviewed. The onboarding tax is real.
19%
Monday Merges
Monday is Merge Day
1 in 5 PRs merge on Monday. Friday deployments? Not as common as you think.
4x
Speed Gap
The Language Speed Hierarchy
Your language choice affects how fast PRs merge. PowerShell is 4x faster than C.
83%
No Review
The 1000-Line Threshold
Once you cross 1000 lines, code review effectively stops. 83% get zero review.
62%→34%
Bot PR Decline
The Rise and Fall of Bot PRs
Bot PRs peaked at 62% in 2022 and dropped to 34% in 2024. What changed?
27%
Weekend Work
The Always-On Engineering Culture
27% of code ships on weekends. 64% of PRs happen after hours. Is your team sustainable?
50x
Variance
What Separates Elite Engineering Teams
From 0 hours to 102 hours—why cycle time varies 50x across top projects.
About Our Research
Data Sources
Our research is based on the GitHub Archive, a project that records the public GitHub timeline. We query this data using Google BigQuery to analyze millions of pull request events.
Why Ungated?
We believe research should be accessible. All our findings are published without email gates or paywalls. If you find our data useful, consider linking back to the original study.
How to Cite
CodePulse. (2025). 3.4 Million PRs, One Uncomfortable Truth. https://codepulsehq.com/research/code-review-study-2025Want to analyze your own team's PR data?
CodePulse connects to your GitHub and shows you these metrics for your repositories.