In his The Information Architecture of Culture presentation at EuroIA in Prague, Martin Belam talked frankly about what worked and what didn't in a recent project at the Guardian.
- The Guardian has 1,373,297 articles on the Internet. 50M users online, which is much bigger than its print audience(~250k). The Guardian has a pretty active community (150k comments a day) and wanted to further engage this audience by allowing them to participate in its art coverage. How can you open up opinions to everyone attending art exhibitions?
- The review/release/consume cycle is out of synch. Previews and reviews come to media outlets before general consumers have had a chance to see/experience things.
- Up front, the team did not get their API model right. They tried to use ISBNS for books and did not heed advice that ISBNs are “evil”. They are a physical system not a digital system. They don’t identify a unique work but a specific edition. They don’t cover anthologies, they are added to CDs, calendars and even card displays.
- Used an Objects/Properties/Actions Map. This helped with mobile first strategy.
- The team created a number of pages that don’t contain original content and ultimately had to de-index them in order to adjust for Google indexing changes that punish sites for non-original content. The Guardian’s reading also thought the automatically created pages were not up to the usual quality standard of the paper.
- When people are arguing about design details, chances are you’ve got agreement on the bigger pictures. Got obsessed with design details (45 minute discussions about start rating details!)
- Make sure there are enough development resources on your team. Lot of people can discuss things but only developers can build. Give developers a chance to be creative. Allow them to hack around your ideas to get more involved.
- The Guardian is now trying two different commenting systems in parallel: one in-house and one Facebook. They suspect the types of comments they receive will be different in the two systems.
- Know what is important: what goal are you trying to achieve?
- Trust good developers. Engaged developers can be the most valuable assets on a project.
- Listen to all of the team. Job titles and age don’t matter if they have the right knowledge.
- Get the model right: lists have been easier than pages and fronts as the model was right.