The Service level expectation graph displays the average time to integrate code reviews or the selected period.
It also displays the code and name of the code review, its cycle time, the distribution of its cycle time, and its integration date. You can display this data by sprint or week, as you prefer. However, note that if a Kanban-type board has been linked to your Axify project, the "sprint" option will not be available in the filters at the top of the page.
Psst! Be careful not to confuse this graph with the one of the same name available in the Process Axis since they do not use the same data (different sources for the two axes).
Reading the graph
For the following graph, the displayed period is the last month, and the display mode is by week. On average, we can see that 85% (85th percentile) of code reviews are integrated in two days or less. As for its equivalent in the Process Axis, we can see that the integration of some code reviews required more time.
The purplish dotted line is the measure used to indicate the 85th percentile of the cycle time of the merged requests over the requested duration.
As this graph is interactive, hovering over an item with your mouse will display more details for that review.
This graph also includes a variation indicator, which compares the average service level expectation for the current period to the average SLE for the previous period. To learn more about the variation indicator and its calculation, check out this article!
Psst! This indicator is handy for spotting outliers, understanding their cause, and fixing them. You can also see how reliable your delivery pipeline is (i.e., if your pull requests are regularly merged in the same amount of days or if it’s somewhat random).
Calculating the metric
The measure visible above the graph (2 days for our example) presents the PR processing time rounded up. Thus, the 85th percentile of 2 days and 1 hour will show 3 days.
The processing time is calculated as follows:
- Retrieve the set of merged pull requests for the last three months.
- Evaluate the cycle time of all pull requests.
- Determine the 85th percentile of the resulting set.