Most of my decade has been spent in a consulting role, which means I end up using the products/techniques/processes mandated by my clients. Occasionally I get to help them improve or change their processes and tools.

To date, I’ve not found “project management” and “requirements management” in the same application / solution. I’ve always had to either juggle two systems, or roll-my-own (usually when the client did not have anything in place and project timelines / politics prohibited introducing something).

I can’t give you a “magic wand” solution, I don’t think it exists (yet). Here’s some of the different stuff I’ve used and my thoughts. Skip to the end if you just want my opinion on what to do. Read through the next section if you want to see stuff I’ve done, which puts my opinions in context.

In the “using products that already exist” camp, I’ve used the following products. I’ve included my “skill level” info, so you can put each opinion in context:

• MS Project : Standard, generic, overly complex for managing dependencies (imho). My skill level: marginally competent. I know enough to believe that there are good ways to use this product. I haven’t had the pleasure of working with / learning from anyone who can do that on a large project (> a few man-years of effort). I’m sure those people are out there.
• Caliber RM : Great schema for managing requirements data. Expensive. Horrible for managing communication. Almost an “input only” tool. I had the most success when pulling a local copy of the database, and generating a static HTML “report” that was really a significant sized hyperlinked website. I was able to create the actionable reports / warnings that were not apparent from the application’s UI (like the impact of schedule delays propagating through traced requirements to affected use cases). My skill level – expert(ish)
• Version Control “solutions” for requirements
  • MKS – expensive. Custom ant scripts made this much better. My skill level – power user on MKS, expertish on ant.
  • VSS – yes, I’m that old. Data corruption issues make this not worth talking about.
  • Sharepoint – ugh. Maybe Sharepoint 2007 will be better, but managing docs (and links) with previous versions was at best clunky. My skill level – competent.
  • Network Share – Versioning files by date, keeping an archive folder, regular backups (both in the network and to my hard drive). Batch files and cron jobs (scheduled tasks) combined with Beyond Compare (great shareware diff-tool) make this workable, but still reliant on people and process. My skill level – power user.
  • subversion – Free, open source, and my hands-down favorite. My skill level – beginner.
  • Lotus Notes – Great when customized. Closed format, only useful when you already have it ingrained in the corporate culture as the source of ‘everything’. Actually evolved some very powerful collaborative databases for in-house teams. Also allowed integration of project management activities (used a Notes DB for that too). Note – this is more than just a repository, it is a working “solution” – but it also served as a repository, so I’m listing it here. My skill level – expert.
• MS Excel – When you know what you need to track, and what you want to report, you can make this work. You’re using a hammer to drive a screw, but you can get it done. I’ve had the most impact/satisfaction when using it to track a work breakdown structure against a set of requirements, including PERT estimates and reconcilliation of estimation quality over time. I’ve also used it effectively to create auto-populated scorecards for reporting project status. If it were less arcane, I would automate linkage of scorecards to powerpoint. Warning: I’ve also been on projects that ABUSED excel as a requirements repository. Change management is all but impossible to do effectively. If you’re managing multiple releases in parallel – forget it. My skill level – expert.
• MS Access – I’ve rolled my own here before too. Once for tracking requirements, and once (more significantly) as a test-automation tool. The effort to benefit ratio is too high to encourage using this. My skill level – competent.
• Visio – I love Visio. Been using it since before Microsoft bought it. I’ve never tried to generate stuff from diagrams, I would look to Lombardi or other vendors to pursue that. I have generated diagrams from source code before (to map dependencies, etc). If using a .NET language, there are solutions that do that for you (and probably for others as well, I was working as a developer in a proprietary language at the time). The ability to create custom shapes and templates makes it a lot easier to standardize diagramming across a project, and get some work-efficiencies for the team. When diagramming processes and building UML models, I go here first. So easy that I don’t try and use rational or other UML tools. The value is in the visual, and it’s easy with visio. Why learn something else? My skill level – power user.

I’ve also rolled-my-own as a C# application, which dabbled in project tracking, but was truly focused on test automation. A bit off-topic, but I would say it is a quick and dirty approach to getting a small team to improve their process. A couple of months of development and we had payback within the first month (on a team with ~ a dozen developers).

Looking forward
I’ve just discovered what might be the ideal software project management solution – devshop. It is a hosted SaaS project management tool that is optimized for software projects. I haven’t used it yet, but I’ve watched all the demos. Here’s my takeaway – looks like it has all the project management stuff I need, and makes leveling/scheduling easy. Not sure exactly how multiple-concurrent-releases would work, but looks very promising. Addresses what I would consider to be the most important elements of risk management. Allows for tying of “SRS level requirements” and designs to specific deliverables. No requirements-specific support. I will absolutely use this for the next project that I can.

For tactical requirements management, honestly, none of the solutions appear to be good enough for me to encourage their use. Someone needs to do for requirements management what devshop appears to have done for (software) project management.

For strategic requirements management / product planning – FeaturePlan may be the way to go. I’m a big fan of Pragmatic Marketing’s approach, and the demo shows what appears to be a compelling approach to managing the process from market research to requirement identification.

I would love to hear what others suggest – especially their opinions on other solutions, and stuff they’ve done to “make it work”. What do y’all think?

Oh – and if you’ve used devshop or FeaturePlan – you must tell us your impression!

Tim