Skip to main content
FeatureFactory
Free tool

Developer Velocity Score

Weight merged PRs by quality, rework and review follow-through to measure durable throughput - not raw output.

The formula

Effort-adjusted velocity discounts raw PR volume by quality, rework and review follow-through: 

score = PRs x (quality / 100) x (1 - rework / 100) x (0.5 + 0.5 x addressed / 100)
  • Quality factor - rewards well-built PRs (0-1x).
  • Rework penalty - subtracts churned/reverted work.
  • Addressed factor - rewards closing review comments (0.5-1x).
  • Durable retention - share of raw PRs that survive as real throughput.

Contributors

Illustrative estimates - edit to match your team.

Effort-adjusted results

Team score: 13.7

Raw PRs merged

20

Effort-adjusted score

13.7

Durable retention

69%

  • Contributor A

    7.1 of 12

    59% durable retention

  • Contributor B

    6.6 of 8

    83% durable retention

Velocity is durable throughput, not raw output

Counting merged PRs is easy - and misleading. A high PR count means little if the work is low-quality, gets reverted, or ships past unresolved review feedback. The Developer Velocity Score adjusts raw output by three signals that predict whether work sticks: PR quality, rework rate and the share of review comments actually addressed. What's left is throughput you can trust.

Read the score as effort-adjusted PRs. A contributor who merges ten clean, low-churn PRs can out-score one who merges twenty that keep coming back. That's the point: the metric rewards work that survives review and stays in production, which is what actually moves a roadmap. Pair it with how to measure developer productivity and DORA metrics for a fuller picture.

Treat every number here as an editable estimate, not a benchmark. When you're ready to track it continuously, explore the Measure product, cross-check output against a PR quality estimator, or browse the full engineering velocity guide.

Frequently asked questions

Adjusted score = PRs merged x (quality / 100) x (1 - rework / 100) x (0.5 + 0.5 x addressed / 100). Quality and addressed review comments raise the score; rework rate discounts it. The result is effort-adjusted throughput rather than a raw PR count.

Measure what you ship

FeatureFactory helps teams plan, build and measure AI web development - tracking velocity, quality and rework automatically so this score updates itself.

Get early access