Presentation given at the 2010 Washington GIS Conference in Tacoma, WA.
Abstract: Software projects have a notoriously low rate of success. This presentation will convey keys to a successful project, gleaned from years of experience both as an in-house developer and consultant. We'll avoid buzzwords and fads, focusing on practical steps you can take to ensure a successful project.
Planning — You can't get anywhere without a map.
Infrastructure — basic tools to keep everything on course whether you're working solo on geoprocessing scripts or with 100 people on an enterprise integration.
Communication — Listen to the right people and develop a decoder ring to understand what they mean.
I got some push back afterwards regarding my comments on estimation, so I'll clarify my thoughts:
Accurate estimation (+-25%) is impossible on any project big enough to be interesting. The reasons for this are two: nothing is as simple as it looks, and good software development is responsive to an evolving understanding of the need.
One exception to my estimation impossibility theory is if you're building something you've already built in the past. If that's what you do...you have my pity and you're probably not even reading this.
Slides at http://www.slideshare.net/kvangork/keys-to-a-successful-software-development-project