Reading/Retrospectives/2014-2015q4

Reading-wide successes

Filling in for PM, team stepped up.
Experimentation in 10% time and hackathon successfully contributing to quarters work and Q1 planning
Engineers taking up PM work has helped keep things moving

Gather
- Built according to goal (share replaced by public feed)
- Saw engagement numbers that surpassed our goals:
- Generic infrastructure around lists re-used in browse prototype, and other types
Wikigrok
- Paused development
- Identified and pushed 120 claims to Wikidata from WikiGrok using very conservative thresholds.
- Used Wikigrok to move people to mobile web beta: now we can validate ideas in beta, more quickly and don’t need to move to stable (2k->43k daily adds, trailing back down, but folks don’t opt out)
Built the Browse prototype in 2 weeks with a minimal number of bugs.
Wikidata descriptors in search results will start rolling out Tuesday, 30th
User testing has been very helpful (ad hoc, heuristic + planned live walk-throughs)
Anonymous editing went live and seems to be succesful
Worked with performance team to identify areas for speed improvement

Incorporated Design more fully into the Development Process
- Begin a more collaborative product/design/engineer design process
- Successfully prototyped link preview feature with design using new process
- Formally integrated Design into the feature and bug signoff for users

Increased Performance of App
- Improved load time of articles and general speed of navigating articles
- Increased perceived performance of the app by improving UX
Improved delivery process
- Implemented and documented a comprehensive release process
- Improved release cadence and tightened feedback loop (5 versions released during the quarter vs 1 release in the previous quarter)
- Automated distribution of "nightly" builds
- Increased unit test coverage by 30%
- Automated running unit tests for builds before automated distribution
- Shipped crash reporting, and used to fix 12 high volume crashing bugs that may have gone unresolved otherwise
- Implemented Regression testing for all deployments to the app store
- Took over ownership of Regression Testing scripts and have added 50 additional test cases (nearly a 100% increase)
- Engaged customers via social media to identify and fix bugs
Contributed to Open Source community
- Multiple contributions to open source Continuous Integration/Deployment project (Fastlane)
- Increased experimentation
- Built 2 prototypes at Hackathon that are being scheduled for development in Q1

Increased unique installs by more than 1.7 million (almost 15% total increase from the end of Q3). Largely due to PR, but owing to new features.
Increased Play Store rating (from 4.36 to 4.39, and climbing)
Continued to demonstrate ability of mobile apps to quickly prototype and deliver new features. (e.g. link previews)

Did not ship
- Lead images (beta, but not stable)
- Browse prototype (not as early as we hoped)
- Site speed (dashboards)
(this impacted Q3 more than Q4) More support for minimizing friction on prototyping--getting extension on gerrit, project on phabricator, required to have documentation
Had moderation communication flare ups with community--need to build bridges there.
Nobody owns developer happiness--vagrant issues are an example here
User feedback loops were slow
Uncertainty around re-org and metrics hurt culture and velocity: we paused development on wikigrok and gather
Passionate people (even within org) telling you you’re doing a bad job can bum you out
Lost engineers to do PM work.

Lead images: We need to engage with the community (and the engineering community) before we start planning. We're more aware now of the cultural and technical problems around lead images (and the infobox).
Browse: Regular review of our beta and stable features is required to stop small projects getting lost in noise, which requires more visibility.

Failed to catch significant "data migration" bugs prior to release (which caused issues for many users and bad reviews)
Failed to deliver link preview feature to end users
Failed to ensure tickets are well defined prior to the beginning of sprints
Failed to improve overall rating in app store
Failed to significantly improve event logging (add search)
Failed to increase developer confidence in making changes to the code base