Failure *Is* an Option
10 likes1,378 views
The document discusses a software project that failed due to scope creep. It began as a project to automate workflows for a proprietary language site, but numerous new requirements were discovered during stakeholder interviews. This led to a massive increase in the scope of the project. Documentation grew out of control, a prototype was never finished, and the functional specification was never signed off. The project cost doubled and the agency lost money while the client lost confidence and brought the project in-house. The key lessons were that application sites require different development methods than static sites, documentation needs management, and scope creep must be controlled.
Failure *Is* an Option
- 1. Learning from Failure or… Failure is an Option @ IA Summit 2007
- 2. What is Failure? The condition or fact of not achieving the desired end or ends. The condition or fact of being insufficient or falling short: a crop failure. Nonperformance of what is requested or expected; omission: failure to report a change of address. A decline in strength or effectiveness. The act or fact of becoming bankrupt or insolvent.
- 3. Usually… “Success has 100 parents, failure is an orphan” John R Ashton (See also similar quotations from JFK, et al.)
- 5. The Case of the Deadly Scope Creep
- 6. The Project A large software company named after a nearby celestial body A consortium site for developing a proprietary language named after coffee A requirement to automate excruciatingly detailed workflows An opportunity to add community and collaborative features
- 7. The Assets A 55-page requirements document - 13 unique roles - 90 detailed requirements - Numerous flowcharts The “constitution” of the community The existing site
- 8. The Discovery Interview multiple stakeholders - Community Participants - Expert Group Representatives - Specification Leads - Executive Committee Representatives - PMO Staff Review of requirements document Analyze analogous, related, and competing sites Present findings and recommendations
- 9. Results of Discovery Scores of new requirements were “discovered” Examples: Logins should be secure Don’t break old URLs/bookmarks Would like to be able to “impersonate” a user Long wish list of reports Blog, message board, wikis and other community features (later abandoned)
- 11. The Wireframes
- 13. The Use Cases
- 14. In Other Words… Documentation HELL
- 15. Long Story Short The functional spec was never signed off The prototype was never finished The IA documents were finalized but too late to save the project The cost of the project more than doubled The agency still lost money on the deal The client lost confidence and took the project back in-house
- 16. Also The site has, nearly a year later, still not re-launched.
- 17. What I Learned Application sites can’t be built with static site methods Documentation is a double-edged sword Scope creep must be managed forthrightly
- 18. thank you Christian Crumlish Yahoo! Pattern Detective Director IT/Web Infrastructure, IA Institute