Answered in reverse order: I'd have done market research first. They all failed because the intended market didn't exist and I couldn't afford to retarget.<p>The ideas:<p>1) Unrealty, a 3D real estate startup. This was tech-driven: Quake 2 and Unreal had recently been released, and I was doing consulting for a real estate developer. Unreal Engine 1 had compelling outdoor scene capabilities, the ability to link areas in a map to and from the web via URLs, and other things. Why not build 3D models of commercial real estate, and have real 3D virtual tours with text and voice chat, AI NPCs who could give "tours" of properties, and link things in the level to web-based stuff like web admin consoles for HVAC systems (or whatever)?<p>So I built all of this. We had some success, mostly due to the novelty. NASA used it for a commercialization demo around the ISS. There was another startup that had us develop even <i>more</i> interesting tech like an in-game web browser and player view tracking for in-game, clickable, ad-served billboards (in 1999!). But, ultimately, it failed because I was not a salesperson, and we didn't hire any salespeople, and architects are generally only interested in doing the amount of work the client needs, and the client (real estate developers) are used to seeing drawings, and the client's clients (tenants) are always going to visit the site in person anyway, so there wasn't really a market.<p>2) Infrastructure for supporting "developer networks," like an MSDN, an API platform community, etc. The problem here is that the infrastructure is the easy part. The hard part is either convincing yourself that you're not a product company any more, you're a middleware company and you have to support your developers really well; or to pay a lot of money to hire dedicated support people (who pretty much have to be really good developers) and then pay them even more money to not go work for your clients once they've learned all there is to know about your middleware or API. If you can do one or the other, you don't need the infrastructure because you'll sort that yourself pretty quick; and if you can't, you don't need the infrastructure, because you don't believe in it.<p>3) Dorm Duffle. Ever go to a Target or a Wal-mart during the weeks leading up to a new semester? The shelves are bare from all the parents and kids buying shampoo and conditioner and soap and notebooks and backpacks and towels and on and on and on. Wouldn't it be nice if all the essentials were waiting for you at your dorm? A big, military-style duffle bag. All you need is your personal effects, books and clothes. Mom and Dad breathe a sigh of relief, and Junior has a sweet college-branded duffle bag to hide the bodies in.<p>Coming from the northeast, that sounds like a pretty good deal, of course I'll pay to have someone else do some boring work for me.<p>But we actually tried it out in the southwest, and got only 1 response. Not 1% from our mailing to the parents of over 15,000 incoming freshmen. 1 single order. Our theory was that in the southwest, culturally, going shopping for kids is <i>something you do</i>, it's important. But I was too broke to follow up.<p>C'est la vie.