A band called the Doubleclicks ran a ridiculously successful Kickstarter campaign in February (over $80K funded of an $18K goal) and wrote up a great -- if long, but long is good for this sort of thing -- article on how to do it. Good stuff, go read it. Thanks to John Scalzi for the link.
One thing I've always wondered about was the stretch goal thing. You know, how we're asking for $20K, but if we reach $25K we'll send everyone who backs us a free T-shirt or something? So you've basically got $5K to spend on T-shirts. That sounds like a lot, but what if you have a lot of people each making small contributions, rather than a smaller number of people each giving more? That's the same amount of money but more shirts owed. What about postage? What about packaging? What if you actually take in $100K, which sounds awesome but if that's another 10K people you have to send T-shirts to, plus whatever higher-level stretch goal premiums you promised...?
Clearly there's a way of making this work -- the whole cost + packaging + shipping thing probably explains why things like wallpapers are so popular as stretch goal premiums -- but just as clearly some people don't think about the details when they design their campaign. The Doubleclicks suggest very thoroughly spreadsheeting everything before you take step one of putting your campaign up. Smart advice; everyone should do it.
Another interesting point was that you need to look at how many active fans you have -- people who already know you, like your work, and who are regularly in touch through a newsletter or your web site or Twitter or whatever you use. That number is a key factor in how much money you can realistically ask for. People who are already popular will be able to raise more money. People who are just starting out and have no friends outside their mother and their office mate at the day job will have a much harder time. Again, it makes sense, but some people seem to think Kickstarter is a magic pot of money that they just have to reach out and take. The Doubleclicks talk about how to analyze the size of your existing network, compare it with successful campaigns run by people in your business and the size of their networks at the time their campaign started, and figure out about how much you can realistically hope to get. Math is good when you're dealing with money.
There's a lot of great info here -- if you've ever thought of doing a crowdfunding campaign, or might some day, go check it out.