Impact Metrics
Open Analytics → Impact (/analytics/impact) to inspect contribution and review activity for the selected repositories and timeframe.
Impact Score
The backend score is bounded to 0–100. Its base calculation combines normalized dimensions:
| Dimension | Weight |
|---|---|
| Change size, with logarithmic dampening | 30% |
| Unique reviewers | 25% |
| Review turnaround | 25% |
| Comment engagement | 20% |
A PR-type multiplier is applied before bounding the result. Contributor totals aggregate PR scores, so a contributor total is not necessarily bounded to 100. The old five-factor 30/25/20/15/10 formula does not describe the current calculator.
What's Displayed
The page combines impact, contributor activity and related velocity/visibility information. High-impact PR listings use qualifying merged PRs. Active team membership can narrow contributor lists within the selected repositories.
Score Calculation Example
Suppose the calculator's normalized dimensions for one PR are:
| Dimension | Normalized value | Weight | Contribution |
|---|---|---|---|
| Change size | 60 | 30% | 18 |
| Unique reviewers | 40 | 25% | 10 |
| Review turnaround | 80 | 25% | 20 |
| Comment engagement | 50 | 20% | 10 |
The base score is 18 + 10 + 20 + 10 = 58. For a PR classified as a feature, the calculator's 1.2 multiplier gives 69.6, before any display rounding. The final PR score is bounded to 0–100. These are illustrative normalized inputs, not a promise that a particular line count produces a score of 60.
Change size and engagement use logarithmic dampening, so doubling lines or comments does not simply double their contribution. Reviewers are counted uniquely, and turnaround depends on timestamped review history. Missing history can therefore change the score independently of the code's value.
Contributor totals aggregate PR scores. For example, three scores of 40, 50 and 60 can sum to 150; the per-PR bound does not cap that total at 100.
Interpreting the Data
Inspect the source change and review context. Score, volume and turnaround describe recorded activity and do not independently establish code quality, employee performance or the business value of a change. Compare the same scope and timeframe, and investigate missing provider history before interpreting an empty result.
Use the listing to find changes worth reading, then inspect the change purpose, affected systems and reviews. A small security fix can have high business value despite a modest activity score. A large, heavily discussed PR can score highly while still needing correctness or maintainability work.

