Stop Wasting Money Writing System Requirements

IT has long been looked to as the technical solution providers for business problems. But the assumption that often is made is that the business problem is already well understood and the business processes that drive the technical requirements are known....by someone. I continually run into the situation where a team is beginning the business requirements definition effort and developing system Use Cases only to discover it's not that straightforward. The entire business process that the technology will support has not been thought through and there are gaps in knowledge around some areas of the business process. How can this be addressed before valuable time and money are wasted spinning on requirements? Here are a few suggestions:

Every IT project should start with business process definition. It's no longer IT's responsibility to only handle the system requirements definition side of the equation. Start thinking in terms of process steps, roles, and responsibilities. Once these are defined the role of the system and how it should behave becomes more clear. The same skill set that it takes to identify system requirements is highly transferable to defining business process. Stakeholder engagement, meeting facilitation, communication skills, clear concise documentation abilities are all the qualities of solid business AND process analyst.

Run an agile project. By adopting the basic principles and philosophy of the agile scrum framework your team can get to "doing" faster with "just enough" documentation in place to make it productive. Once the business process is well understood, create a backlog with the goals the system should allow each type of system user to achieve. Define requirements in more detail as part of each development sprint. Start the sprint with a few upfront requirements tasks then continue to refine the requirements as a team through the build-demo-adapt process. The role of the BA in the demo meetings is to capture requirements and decisions made. By the end of each sprint you can have a requirements document that matches exactly what was built. This avoids spending months in lengthy requirements gathering sessions trying to predict in excruciating detail today how you want the system to work a few months from now.

These are a few ideas that have worked well on my projects. Join the conversation! What are you doing that's worked well(or not so well)?