Wikimedia Discovery/Meetings/Analysis retrospective 2016-03-17

From MediaWiki.org
Jump to navigation Jump to search

Covering whatever has happened related to the team since the last retro (2016-02-22)

  • Oliver is leaving
  • Lila left; Katherine interim ED
  • Completion suggester rolled out to production
  • First portal improvements rolled out to production
  • Dashboard improvements rolled out
    • Browser breakdown
    • Language/project pair breakdown
    • External traffic to Portal
  • Deb asked for a lot of small improvements to the dashboard (see ^^)
  • Codebase refactored (thanks, Oliver!)
  • Reports for Portal team: JS support, 10% referred traffic
  • Published data access guidelines
  • Published Discovery Analytics Employee Operations Manual

Review action items from before[edit]

  • n/a

What went well?[edit]

  • Hiring delegation (working on JD and task). As noted, Dan delegated to Mikhail simply by thinking "I don't have time for this! Oh well." and Mikhail read his mind.
  • Dashboard improvements are awesome (thanks! :D)
  • Analysis pages up on meta
  • We did an A/B test that showed the completion suggester reduces the zero results rate, we pushed it out to production, and guess what, it reduced the zero results rate. Isn't that neat?

What could we improve?[edit]

  • Still seems to be some confusion? or difficulty? creating the suite of tasks related to an A/B test
    • Clarification on this? Dan experiences no such confusion.
    • Kevin acknowledges that this might actually not be a problem at all--could be a mis-perception
  • Hiring Oliver's replacement. Dan is bogged down with work and hasn't had the time to devote to making it move.
  • Took a while for the data access guidelines to be approved by legal (part of overall pattern of legal OR OTHER PEOPLE outside the team taking forever to reply, requiring lots of pinging)
  • We did an A/B test that showed the portal improvements increases the clickthrough, we pushed it out to production... and it didn't seem to increase the clickthrough. That's confusing. (https://phabricator.wikimedia.org/T130235)

What else should be noted?[edit]

  • About to try new backlog workboard structure
  • Want to be sure we don't overload Mikhail with too many analysis tasks from the Portal team
  • Mikhail feeling a little bit of a meeting/context-switching overload, but doesn't think it's a big problem just yet.
  • and once again, Oliver is leaving :(
  • At times, analysis hasn't had tests to analyze---lumpy workload
    • But it did give the team time to do refactoring and other valuable but non-urgent work
  • We don't have an outward-facing analysis page (like we do for maps etc). 

Retro of retro[edit]

  • Kevin learned some about time management (half an hour of retro goes quickly)
  • Mikhail is happy that we ended up with a full retro doc, reflecting the impact of this team

Action Items[edit]

  • Chris: We should consider publicizing the data access guidelines more widely
  • Chris: We should probably create an outward-facing analysis page
  • Mikhail: A task was already created to try to answer the portal clickthrough rate discrepancy (https://phabricator.wikimedia.org/T130235)
  • Kevin: Discuss legal responsiveness w/TPG to see if there are wider org issues