A change can take days to deliver while receiving only hours of active work. CPU time here means activity time, also called touch time: the time spent directly progressing a work item. It includes coding, writing or running tests, debugging, addressing review comments, reviewing, merging, and deploying. It does not mean machine CPU usage. section 5.4.2 section 9.2.3
Exclude time when the change is merely waiting: a review queue, a CI result or rerun, a deploy window, or an idle handoff. Count an automated step when it directly transforms the work, but do not count an entire pipeline run as active time if most of it is blocked or queued. This distinction matters because lead time and cycle time show how long delivery took; activity time helps reveal how much of that duration involved progress. section 5.4.2 section 9.2.3
Table 5.2 Activity ratio analysis of a simple workflow
|
Stage
|
Activity time
|
Waiting time
|
Cycle time
|
Activity ratio
|
|
Implement change
|
30
|
0
|
30
|
1.00
|
|
Run local tests
|
1
|
0
|
1
|
1.00
|
|
CI pipeline (1st run)
|
1
|
7
|
8
|
0.13
|
|
CI rerun (2nd run)
|
1
|
7
|
8
|
0.13
|
|
Code review
|
6
|
234
|
240
|
0.03
|
|
Merge
|
1
|
0
|
1
|
1.00
|
|
Totals
|
40
|
248
|
288
|
0.14
|
The workflow totals 40 units of activity time across 288 units of cycle time, for an activity ratio of 0.14. Review accounts for 6 active units but 234 waiting units. Making coding faster would not remove that delay. section 5.4.2
A review queue, before and after
Suppose review takes three hours of active work but the change spends four days at the review station. Dedicated review windows reduce that station’s elapsed time to 1.5 days while active review remains three hours. The improvement comes from a shorter queue, not harder work. section 9.2.3
To use the metric on a real change:
- Select a recent completed change. Reconstruct its path from first edit to merge, estimating time spent coding, testing, fixing, and reviewing separately from waiting. section 5.8
- Compare activity time with elapsed time. Activity time divided by cycle time gives a within-work activity ratio; an exercise measuring against the full request-to-delivery lead time instead includes earlier backlog waiting. State which denominator you use. section 5.4.2 section 9.5
- Find the longest delay, change one cause—such as a flaky test or slow CI stage—and measure the next change. A low ratio points toward waiting; it does not, by itself, identify the bottleneck. section 5.8 section 6.4





