Get the new 2025 DORA State of AI-Assisted Software Development report →
Changelog · Feb 15, 2021

Pull request cycle time trendline and breakdown by stage

A software team’s velocity is a combination of many factors. Looking into pull request pipeline by stage is a great starting point to identify bottlenecks and get pull requests through faster. With the new cycle time breakdown it is easy to see whether review, merge, or WIP time takes longer than expected.

You can also use Swarmia’s filters to look at cycle time and throughput metrics for bot-created, cross-team, and other PRs separately.

The chart below cycle time shows how your pull request queue trends over time. When multiple pull requests are left open for long, it trends up. When pull requests are merged timely, it trends down. Many teams will notice that Christmas bump on longer timeframes where cycle time increases during holiday time 🎅.

Pull request age now includes the time since the first commit

first commit

Previously, pull request age was calculated from the moment a pull request was created, and earlier commits would not get reflected in cycle time insights. Including time since the first commit allows better comparison between pull requests and makes pull request insights more accurate.

Other improvements

  • Throughput (pull requests merged per day) chart got updated, and now includes the total number of pull requests merged in the selected timeframe in addition to the daily average.
  • Cycle time and throughput metrics now include a comparison to the previous time period

Subscribe to our newsletter

Get the latest product updates and #goodreads delivered to your inbox once a month.

More changelog updates
· Nov 3, 2025

Better effort tracking for development-adjacent work

Your engineering team does more than just write code. Bug investigations, planning sessions, testing, and other development-adjacent work are crucial parts of the software development process…
Read more
· Oct 27, 2025

Track Claude Code adoption and use in Swarmia

Introducing AI tools like Claude Code is one thing — getting your teams to actually use them is another. To help you bridge that gap, you can now track Claude Code adoption and activity…
Read more