Experiment with timezone-based roulette suggestions

In gitlab-org/gitlab!34862 (merged), we're replacing the current random-based reviewer roulette suggestions with timezone-based suggestions.

The goal is to reduce the mean time to review and mean time review to merge for the GitLab project by using timezones for the the reviewer selection process.

Feedback and results will be posted to this issue.

Experiment duration

We'll run the experiment for at least 1 month before taking a decision to keep the new strategy or revert to the old one, unless it's clear that the new strategy is not better than the current one.

The experiment started approximately on 2020-07-08 and will finish approximately on 2020-08-08.

Expected outcome

  1. P50 days to ready for review is consistently less than 0.02 days: https://app.periscopedata.com/app/gitlab/647974/WIP:-R%C3%A9my-Coutable-Scratchpad?widget=9058640&udv=1002157
  2. P50 days for first reviewer response is consistently less than 0.25 days: https://app.periscopedata.com/app/gitlab/647974/WIP:-R%C3%A9my-Coutable-Scratchpad?widget=9061564&udv=1002157
  3. P50 days from review to merge is consistently less than 1.3 days: https://app.periscopedata.com/app/gitlab/647974/WIP:-R%C3%A9my-Coutable-Scratchpad?widget=9061224&udv=1002157

Feedback

Feedback is welcome at gitlab-org/gitlab#227123 (closed).

Edited Jul 09, 2020 by Rémy Coutable
Assignee Loading
Time tracking Loading