Once again the timeless developer question: what's your methodology? The question came up tonight, here's my current answer:
The shortest path between the current situation and the goal. And SharePoint projects are completely different from software development projects, except for the software development parts.
What I do is a mix of:
- Motion - always study the existing situation and plan the solution in terms of capabilities, not features. If you follow this route, avoid Motion "Lite".
- What Jim McCarthy describes in Software For Your Head. This book explains how to get teams working with total clarity in the same direction.
- Personae and Scenarios. Again back to the capabilities. These are described in MSF4 but really it goes back to Inmates Are Running the Asylum.
- Visio diagrams in any form that can be followed by an average person coming in cold. If people don't get it, draw it another way and provide that too. UML is consistent but as it turns out projects aren't. What I mean is that it's a great visual language in the same way that C# is a great development language. If you invest the time in speaking it, it's a great way to explain things to others who speak it. Unfortunately they don't teach it in business or arts courses so it tends to be wasted on decision makers. Edward Tufte has some great books and webs on presenting visuals, unfortunately there's no "how to," it's again "practice until you're good."
People whose skills I admire got that way the same way they do in any field: read and try a whole lot, internalize the methods, and then forget the methods and simply express. The best style is to have no style. Once you have enough information there are no choices, only what is to be done. It worked for the greatest distiller of philosophy to live in the 20th century, it can work for you.
Or roughly, with portals and collabortive sites it's X-ish methods minus the crap about the little cards and 10-minute stand-up meetings. Walk away for a week and they'll stop happening, and that smell bad enough to think twice. The rest - tracer bullets, many iterative milestones, the build is the demo, continuous engagement, heck, I guess I'm really talking about Code Complete. Get it, it's a gooder.