There’s a prominent Atlanta technology company that launched a new, full rewrite of their product this year. It was a disaster. I’ve been there as well. In fact, I know of several other software company CEOs who have similar battle scars.
A simple piece of advice: never rewrite a product from the ground up if you have a considerable customer base.
Here’s what I learned when we did a full rewrite four years ago:
- It literally will take twice as long as your worst case scenario
- Morale will be significantly affected on the engineering team
- Sales will suffer as reps continually think the new product is around the corner
- There will be major bugs upon roll-out, even with significant product QA
Now, we’ve all heard these issues happen to normal large IT projects, but not to software companies that specialize in this, right? Trust me, it doesn’t matter. People are overly optimistic that things get done in a shorter amount of time. My recommendation is to plan the big rewrite but to implement changes incrementally, instead of with a single release.
David Cummings on Startups 
I’ve done it, too. Twice. For effectively the same product. It is amazing we made it through. It cost us a lot, though. Sometime I ought to write a bit about the experience.
In the meantime, Joel Spolsky and Chad Fowler both have good posts on the topic:
http://www.joelonsoftware.com/articles/fog0000000069.html
http://chadfowler.com/2006/12/27/the-big-rewrite
Thanks Coty for the great links. Two separate people brought it up to me today that the Netscape software rewrite was considered the death knell for the company.
Hear! Hear!