Web O’ Wonder Release Retrospective 6/3

Planning

• Initially thought that the site would live forever
• The vision changed a few times
• We were never certain how long it was going to last
• Having long term plan decided at kickoff would have saved a lot of development time.
• We were unsure where to host demos
• Project happened around same time as web stewards process started, so it wasn’t fully baked yet.

UX/Design

• Didn’t have a very clear idea of how many demos we were going to have
• Initially thought UX design would scale well to 30 demos, but would have made different decisions based on the lesser amount we ended up with.

Staffing

• Working with weight shift went well. We liked working with them--they knew their stuff
• Weightshift was originally contracted to design the site
• Forgot to think of mobile design. Luckily Jason jumped in and did awesome job
• Miscommunication on who was going to build the site.
  • Weightshift jumped in and did front end and Moz did back end.

Schedule

• Always on a tight schedule, but overall it was realistic and went well.
  • Nov kickoff
  • Jan commit code
  • Push March
• Always knew this was Fx4 dependent
  • If Fx4 launched in Feb would have been rough
  • Fx4 launch coincided well with WoW

IT/deployment

Corey’s team was pushed to the limit during Fx4 deployment. They pulled it off thankfully, but cramming too many campaigns into a small window of time was too much. Shouldn’t be a problem going forward with rapid release schedule.

WoW deployment went well considering core IT priorities were web properties supporting Fx4 product and downloads: AMO, input, etc. That’s why things seemed hectic.

Paul and demo authors

• We hit scaling problems due to backend component.
• Fixed it going forward to client side only
• Have to be careful with third party development

WoW is different.

• Not a dynamic site.
• Not collecting user data.
• All static content
• not a lot of backend compents
• Only locale detection, mobile detection
• This was the first playdoh app
• Twitter party was the problem
• Dark launch helped

Localization

The plan was English only to start but ran into constraints when beginning l10n. Had we decided to localize from beginning would have been OK. Now we do l10n on all projects.

Stas:

• Overall it went smoothly
• Could we dark launch and have localizers test as well?
• We could let localizers into stage.
• Some CSS and sizing issues with design
• Had to fix navigation bar on a few locales
• Stas working on document to test early on for sizing issues
• Web fonts gave l10n some trouble
• Hoping to create public domain font face.
  • Currently work well for western language but break badly for slovek, asian languages
• Juggling between stage, locales and production kind of a pain.
  • fixing this in pladoh
• URL scheme we use for websites in general worked well for WoW.

Communication

• Demo designers -- Paul was there to help them.
• Laura was there to do some PM, getting them to do things on time.
• Expected them to lag, but they did good job at keeping on schedule
• Would be harder if they were volunteers/contributors
• IRC channel was helpful
• Laura and ozten lots of communication
• All contributors were motivated and excited

Metrics

• Use web trends or StatsD for better tracking.
• Needed more robust tracking. Didn’t know which demos did the best.
• Twitter and social analysis showed which ones did best.
• Wish we could see how much design made an impact.
• Facebook and twitter usually an afterthought.
  • Have we studied what value these additions provide?
  • Would like to take control of these buttons instead of just dropping them in.