OK, here's what works for me:
1 - pick something related to your topic - keyword-based is my preference, but anything related, shorter the better
2 - if it's not available - and it likely won't be - start looking in places like www tdnam com - do a search for expiring
domain names with a keyword your after in it and restrict the search to .com domains only
3 - if you see one you like, bid on it
4 - if not expressly searching under expiring domains
, you could find what you're after and just buy it immediately, too
I usually check the following points on any domain before purchasing it:
1 - look through Yahoo (links: insertdomainhere) to get a feel for any inbound links that might already exist to the domain
2 - check the Yahoo directory to see if maybe it's already in there - bonus if it is, no biggie otherwise
3 - I use archiveDOTorg to see if there is any record of the old pages on the domain - if there is, it'll give you a good idea of the focus of the last version of the site - this can help determine if any inbound links will ultimately have any value for you moving forward.
4 - I scoot through DMOZDOT
org to see, if by some miracle, there's a listing - but I don't sweat it if not...
5 - finally, I'll use a Page Rank lookup tool to see if there's any PR still hovering around the domain
I also try to visit the domain directly, and do searches on it through Google to see what they might have showing for it, but this is more "curiosity" stuff for me...
I have bought a few domains this way and managed to resurrect one and it now actually turns some Adsense revenue and so far (5 months) has managed to keep it's old PR values intact, too.
expiring domains with inbound links, PR values and listings in desirable places will usually be MUCH more expensive than just buying one you invent and is available. The rewards *can* outweigh the extra cost, but it's a balancing act with risk, so tread lightly.
In the end, make sure the domain is relative to the topic and easy to recall.