Wikimedia Discovery/Meetings/Search team health check 2016-11-30


The team health check survey was developed by the Team Practices Group as a way of allowing teams to self-assess their health. This session modified the original format to avoid forcing the team to come up with a single consensus rating for each focus area. Instead, the summary for each focus area attempts to capture the gist of the discussion, along with any notable points, some of which could contradict the summary.



  • Internal communication within the team is very strong. Members have a high level of trust, and appreciate being "bugged" with reminders or questions. External communications seem fine, with healthy ongoing experimentation and learning about how best to interact with communities.

Mission and goalsEdit

  • Clear ties from our org vision and mission to the work of the team. Some questions about org strategy, but despite that and other turmoil, this team has worked very productively. Some team goals might be high-level, but the team turns them into concrete, valuable work.


  • The team feels very empowered as a group. The PMs work with both community and engineers to set priorities.


  • The team feels very supported by the organization. Having our own analysis, ops, and UX/design people reduces external dependencies. Embedded CL and TPG have few if any pain points.

Community involvementEdit

  • The team has a good relationship with the user communities (CL is key to that), although at times it would be nice to get more feedback. We have built trust through planning and execution. We don't get many technical contributions, probably due to the specialized and back-end nature of much of our work so far.


Solid. No major breakages. Overall, tech debt is decreasing. Code review is effective, and automated tests are improving. Front-end quality should improve as Jan gets up to speed.


  • Metrics and anecdotes both indicate that search is getting better. We hear from people that it is improving.


  • Mostly excellent. Some room for improvement with the wmf config stuff and reindexing.


The work is complex and involves some ancient code, which means it can't move as quickly as one might hope. Some of the challenges are inherent, but some could be improved. The UI should move faster as Jan gets familiar with the code. A/B tests can take up to 2 months to design, run, and analyze. Work hours seem sustainable.


  • The work is tricky and has a high learning curve, so people are learning a lot all the time. There is always more to learn. We also experiment with process changes.


  • The team has a lot of fun together. Everyone is knowledgeable, interesting, and has a generally positive attitude.


  • This was an interesting alternative to a retrospective, and it covered some new topics
  • It is time-consuming, so probably only worth doing maybe annually
  • Teams with issues would probably benefit from doing it more often


  • Dan: work with apps team on full-text searching thing
  • Dan: helping slow moving UI work move faster