Agaetis — Studio Analytics

A portfolio dashboard for a fictional agency. All data is invented.

Median Cycle Time

27.2h

↓ 21.2% YoY↓ 3.9% MoM

PRs Merged

↑ 42.1% YoY

Change Failure Rate

4.5%

↓ 1.0pp MoM

01Cycle Time

Cycle time · By phase

Hours to merge

CodingReview wait43% of the clockMerge

Efficiency

27.2h

↓ 21% YoY

CodingReview wait43% of the clockMerge

Cycle time · Where it goes

Active vs waiting

Median PR in August 2026, by where the clock ran

Retention

57%

Of the clock was spent on active work

Median PR in August 2026, by where the clock ran

02Throughput

Throughput · Monthly

Pull requests merged

PRs mergedMedian cycle time

Growth

27.2h

Median cycle time · 81 prs merged

PRs mergedMedian cycle time

Throughput · Movement

PR movement

Opened78Reopened0Resolved82

Growth

-4

Net last month · 1.6× avg

Opened78Reopened0Resolved82

03Release Cadence

Release · Cadence

Deploys vs target

GoalActual

Efficiency

Deploys · 104% to goal

GoalActual

Release · Quality

Cadence + reliability

Efficiency

103.8%

Cadence attainment · down 14.0pp from peak

Cadence attainment103.8%Deploy success95.5%

04What We Build

Engineering · Work mix

What we build

Merged pull requests by type · Apr–Aug 2026

Retention

34%

Top type · Feature

Merged pull requests by type · Apr–Aug 2026

Stacked bars show the phase breakdown for each month. Compare the total with review waiting to see where elapsed time changes. This measure ends at merge, before deployment.

Review waiting peaks during the launch bottleneck while coding time changes less. The later decline in waiting coincides with higher throughput and more on-time launches.

The active slice combines coding and merge phases. The waiting slice shows review wait. The centre shows total hours; the legend shows each slice's hours and share.

Waiting remains a substantial share of cycle time after the queue improves. Compare it with the phase history to distinguish the latest split from the peak backlog.

Bars show merged PR counts; the line shows median cycle time in hours. The measures use separate scales. Compare their trends, not the line's height against a bar.

Higher throughput alongside shorter cycle time can indicate a queue clearing. PR size and complexity are not shown, so counts alone do not measure productivity.

Opened PRs add to the queue. Resolved PRs combine merges and closures without merging. Their difference is the backlog change. Earlier work can finish this month.

Positive net values mean the backlog grew; negative values mean it shrank. Compare the pattern with review waiting and merge throughput.

Paired bars compare production deploys with the target for each month. Targets can vary. Hover for target attainment and the difference in deploy count.

Compare cadence with deploy success to assess frequency alongside reliability. A higher deploy count alone does not show whether releases needed corrective work.

The first line divides actual deploys by planned deploys. The second shows deploy success rate. The 100% reference means plan met for cadence, or all deploys successful for reliability.

Catch-up releases lift cadence above plan after the bottleneck. Compare deploy success alongside that change to assess reliability.

Slices group merged PRs by work type over the stated window. The legend ranks types by count. Each PR has one type, regardless of its size or effort.

Compare feature work with fixes, infrastructure, and maintenance. These are shares of PR counts, not shares of engineering time.