Blogging about a project just before its go-live date, especially when it is centred around a single live event, feels a little scary - I hope we are not jinxing it by doing so. It’s however worth doing it now because this is something of an experiment and a learning experience, so what better time to share than when you’re sprinting for the finish line?
The experiment is about using Facebook’s Places in the context of a live event - in this case Radio 1’s Big Weekend. There’s more background on the project on the BBC’s blog.
The process that got us there was an interesting one. Using Facebook Places via the new API and checking into gigs rather than physical places wasn’t something we had done before or could research, given that I guess very few outside of Facebook would have had the chance to play with it. We still had to commit to an unmovable deadline and a fixed budget.
We got there by being truly agile and thanks to the BBC’s team’s readiness to trust us with it.
So what do I mean when I say truly agile? We have been running Scrum for a few years, so this project was no different to many others we have run: a backlog, sprints, daily standups, iterative development. It however worked out particularly well because of the way we prioritised our work:
- We started by developing the core functionality - getting line-up data into the system and talking to Facebook’s APIs
- Styling and copy were applied slightly later on and evolved through the project
- We identified features (stories in Scrum talk) that the BBC could have done without, if it came to it, so we got to a release candidate a lot earlier than planned
- While we ended up delivering most of those, having a minimum spec web app ready to release did mean that we were able to slot in higher priority and unexpected tasks into the project without taking the risk of completely missing our deadlines
The most significant ‘unexpected task’ was related to the check in process. Facebook users are effectively checking into an artist URL that also includes Facebook metadata. The original plan was to point to the BBC’s website URLs for each artist. As the project progressed, it became clear that coordinating the necessary changes on the website would have taken too long. We therefore ended up implementing a more complex system of redirects that does however give us the advantage of better tracking and more control over the user experience.
Was it all smooth and perfect? Of course not. We hadn’t appreciated the level of detail required to satisfy the BBC’s information security requirements; if we had, we would have planned for more upfront technical architecture work. While all parties showed plenty of flexibility, we were still operating under a fixed price contract, so we had the inevitable tense conversations over change requests vs. expected features.
All in all though, we are very happy with what we have done and we are looking forward to the weekend to watch it all play out. So particular thanks to Richard and Robert at the BBC, much appreciation for Simon’s support at Facebook (great dev support guys make a massive difference when working on new platforms) and well done to the team: Jon Markwell & Chris Williams on development, Aegir Hallmundur on design, Paul Welsh on QA and Cori Samuel keeping it all running smoothly.
Comments