Program for yourself, not others
This has been said many times before, but I think it really sunk in for me today. The best software business you can start is one that solves YOUR problem. Just today, someone came to me with a "great idea" and the first thing I asked is:
"Well, would you use something like that?"
"Um...no, but I think a lot of people would really.."
That's the problem. If you can't convince yourself to buy whatever your selling, it's going to be hard to create something to sell to fictional future customers. "But I asked my mom and my brother what they thought, and they said it's a million dollar idea". Here's the thing. Agreeing to purchase fictional software is easy. Actually doing it is another thing. That's why market research can be so difficult. People lie. Maybe they want to impress you, maybe they just don't want to be disagreeable, but it's almost impossible to gauge actual future sales simply from asking people if they want whatever you plan to build.
If you look at the successful ISV's out there, almost all of them sell MeWare, as Eric Sink calls it. It's a no lose situation. If you solve your own problem, even with no sales you have something you can use. But there is a good chance other people have that same problem but are lazier or not as creative as you.
When I created my budget program or my used car site, I knew exactly what I wanted, exactly what would work well, and as a result these are both fairly popular. ChimSoft on the other hand is something I designed for others. It was a lot harder to sell version 1.0 because I had one person's idea of what they wanted, and then a few others ideas, but since I personally wouldn't truly use the software, I had to make tough decisions as to what the people actually wanted.
If you don't use your backup solution, or you don't use your handy dandy band-tracking RSS feed, it will be unlikely to work in a way that is useful to others. Marketing can make up for this in some respects, but most likely it will take a couple of versions with real users feedback to get a quality product.
How does this all relate to me? I've decided to try and release the Bug/Feature/CRM/Mailer tracking thing I discussed in a previous post, because this is something that solved all my problems. I know how it needs to work, and even if no one uses it, it's still good for me. And if anyone can think of an ISV that is sucessful without creating me-ware let me know, I'm really racking my brain.