Metrics explained

Understanding your numbers

Every metric in Scryable is derived from your git history. None of them require you to install anything or change how your team works. This page explains what each number means, how it's calculated, and what to look for when comparing current figures against your pre-AI baseline.

Commits

Total commits

The number of commits made within the selected date range across the selected repositories and branches. Merge commits are included. Bot commits are excluded where they can be identified by author email patterns.

On its own, commit count tells you relatively little. A rise in commits post-AI-adoption is common and expected. What matters more is whether other quality signals are moving in the right direction at the same time. A team making more commits with stable or improving churn is in a good position. More commits with rising churn is worth investigating.

Active contributors

The number of distinct commit authors who have committed at least once within the selected date range. This is a headcount signal, not a quality signal. It's most useful for confirming that your filters are set correctly and you're looking at the expected set of people.

Line changes

Additions

The total number of lines added across all commits in the selected range. This includes new files, new functions, and edits that increase file length.

Deletions

The total number of lines removed across all commits in the selected range. Deletions include line removals, file deletions, and edits that reduce file length.

Gross lines

Additions plus deletions. The total amount of code activity, regardless of direction. High gross figures relative to net figures suggest a lot of rewriting rather than net-new work.

Net lines

Additions minus deletions. The actual growth of the codebase over the period. A team adding 10,000 lines and deleting 9,500 has a net of +500. Net is often a poor proxy for productivity on its own, since a refactor that reduces code size without reducing capability is genuinely good work.

Large outlier commits (a single commit adding 50,000 lines from a vendor library, for example) can significantly distort these figures. The outlier filter in the sidebar lets you exclude commits at the P95 or P99 level to get a cleaner picture of typical activity.

Churn ratio

Churn

Churn measures how much recently-written code gets rewritten or deleted shortly after it was committed. It's the metric most closely associated with AI-assisted code quality.

A churn ratio of 1.0 means code is being deleted at the same rate it's being written. A ratio below 1.0 means the codebase is growing healthily. A ratio above 1.5 is a signal worth paying attention to, particularly if it has risen since your pre-AI baseline date.

Churn is a signal, not a verdict. High churn can mean AI-generated code that needed immediate correction. It can also mean a team in the middle of a deliberate refactor, or a project in early-stage exploration. Context matters. Compare the current churn ratio against the pre-AI baseline before drawing conclusions, and look at who is driving elevated churn before deciding what it means.

Research from GitClear (2025) found that AI-assisted commits show an average of 2x higher churn rates compared to human-authored commits. Scryable lets you verify whether that pattern holds for your specific team and tooling.

Burst

Burst measures the clustering of commit activity. A developer committing steadily throughout the working day has a low burst score. A developer who goes quiet for long stretches and then pushes a large volume of commits in a short window has a high burst score.

AI coding tools tend to increase burst because they accelerate the output phase of development. A rise in burst post-baseline is common and not inherently a concern. Extremely high burst scores can indicate commits that haven't been reviewed before being pushed, which is worth watching if churn is also elevated for the same contributor.

Density

Density is the percentage of working days within the selected range on which a contributor made at least one commit. A density of 82% means commits on roughly 4 out of every 5 working days.

Higher density generally indicates consistent day-to-day progress rather than large batch pushes. It's most meaningful as a baseline comparison: if a contributor's density has risen significantly since AI adoption, it may indicate they're committing smaller, more frequent units of work.

Active days

The raw count of days with at least one commit, shown as a fraction of working days in the period (for example, 18/22). This is the absolute figure underlying the density percentage. It's shown at the author level in the deep-dive view.

Weekend %

The percentage of a contributor's commits that were made on Saturday or Sunday. Scryable surfaces this not as a productivity metric but as a wellbeing signal. A significant rise in weekend commits post-AI-adoption can indicate that AI tools are creating pressure to deliver more, including outside working hours.

This metric is intentionally presented without a "good" or "bad" framing. Some teams work flexibly and weekend commits are unremarkable. Others will find a rising weekend % worth a conversation.