Over the years I've started a small handful of Internet/software projects/companies. Examples include Mailinator, Preemptive Solutions, Inc., and Classhat. Actually, I've started a large handful but no one knows about most of them because they were (in no particular order): dumb ideas, unsuccessful, too hard for me to complete. Given that I now rate any new idea I get according to a set of rules that helps me filter out good ideas from bad. At least, whatever I consider bad. Keep in mind these rules are for the canonical one or 2-person pre-startup - if you have 8million in VC, there's a lot of other magic you can do.
Here they are:
1) If there is no business model, its a hobby, not an idea. I love compiler optimizations. I wrote a Java optimizer soon after Java came out. I spent months trying to figure out how to turn it into a business. But guess what, people don't pay for optimizers, or compilers, or even runtimes. At least not without a strong sales team telling they need that. By and large most ideas I get are about things that I'd love to work on but have no real business model (ala my Classhat project took several years and is, absolutely, a hobby). There's nothing wrong with hobbies, as long as you know what they are.
2) The best ideas make your customers money. If your idea can say "If a customer uses our product, they will make X% more money" (where X is a positive number, even if quite small) - you have won the game. Importantly - I did not say the customer will save X% more money. I said they'll make it. That's a big difference. Saving money is great, but you are then faced with the mission of convincing your customer that if they spend $100k on your product now, it will pay itself back in 8 months. It's way way way easier to say "Use our product and you make 2% more money (of which we get a cut). Don't use it, and don't." Who wouldn't buy that?
3) The best place to be that I know of is B2B2C. That is, you want to be a business that serves businesses that serve consumers. If you're B2C, then welcome to some important challenges. One is to get people to pay for your stuff, which in this Internet world, we're not all that happy about doing. Secondly, welcome to support hell. Its very hard to provide consumer support (and you see many complaints across the net). It takes a lot of support people and a lot of money to do it right, which is why it rarely is. If your idea plans to charge consumers, I'd definitely think twice unless you can ramp up a support system fast. Thirdly, you'll need a powerful infrastructure (apart from support) just to handle large numbers of small transactions. Its harder to sell 1000 $10 widgets than 10 $1000 ones.
On the other hand, if you're simply B2B, life isn't so bad. Big ticket sales but encompassing the market is harder. Sales cycles are long. Again, you might need a faster ramp-up of a sales and marketing infrastructure (which you are going to need eventually anyway), but you're probably ok if the idea is generally good.
4) If you're going B2C, look for revenue models that don't come right from the consumer. Given the last point you're probably thinking I'm crazy given things like youtube or myspace or facebook or google. I'm not (at least I don't think I am). All those places dodged the problems of providing support and tracking sales by giving away their services for free and making money on the backside (whatever that is). Often that's advertising revenue. Often its partnership deals. Simply put there absolutely is an Eyeball Business Model. If you can get the eyeballs, you can sell them. Just try to do that instead of charging them directly. They'll be ornery about it and demand support.
5) Revisit every bad idea every once in awhile. Why is AJAX hot? Because its enabling things technically that weren't enabled before. Are any of these mashups really novel ideas? Not usually. Old ideas become new again by new technology, faster computers, etc.
When the computer game Doom came out it included very little novel technology. All the 3D math and graphics tech had existed for years. The guys at ID were just the first ones to realize that personal computers (as opposed to graphics workstations) at the time had finally gotten powerful enough to do it all in realtime.
Every time bandwidth gets faster for cheaper, previously bad old ideas become new and shiny (e.g., video over the Internet). Mobile phones seem really ripe. Apps and usability currently sort of sucks. Thats really hampering a ton of uses. With every new advancement however, its going to open up new doors.
6) Do your best to create a system of recurring revenue. Advertising on websites and such is easy. But things like packaged software is hard. Why does Microsoft change the doc format every time it releases a new Microsoft Word. Surely it probably includes new features, but why can't old versions just ignore those and load the file (which is really text right?) anyway? Because they don't want it to. They want you to buy a new version of Word to get all the new features (even if you just write letters to your Aunt Edna and would never use them). They want recurring revenue.
Doing that with packaged software is harder and often somewhat transparently greedy. Heard of "software as a service"? That is of course, really, "software that we can just keep charging for". Bottom line is if tires, light bulbs, or razors never wore out, the economics of those businesses would be radically different. Getting more money from your customers (hopefully while providing value to justify it) is a good way to go.
If have an idea that follows #2 above (making your customer money), do your best to simply take a small cut. Small cuts add up and the customer has no risk in trying your product.
Edit: There is now a part TWO to this article.