I usually come up with names after something I see or hear, sometimes swapping letters around or adding 1-2 letters, sometimes replacing letters. I have also “discovered” names, serendipitously, by making typos when I’ve been typing something, like an email or a url. It is surprising to me how often this happens and then my workflow stops as I look at and think about what is on the screen. Sometimes, its nothing or really just junk, but sometimes it looks and sounds great to me. I am not into registering any keyword domains and never have been.
For almost every domain that I think about registering, I enter it into my s/sheet and then leave it for a couple of days. I also speak out the name to see how it sounds. I then go back to it after a couple of days and see how I feel about it then. Only after I think its still a good name do I then begin the process of seeing if the name is available and if there are any claims to the name, trademarks, similarly sounding or spelled names, etc.
Usually, I need to like how the domain sounds and if it passes the “radio test”. I also have to be able to see what market the name might address or appeal to and whether that market is immature/mature and/or overcrowded and/or populated by upstarts. Domain names are very low cost, but if you register many iffy names or names with fringe appeal, the costs add up.
I check the visitor figures for each of my domains as they are coming up for renewal and take a decision on whether to renew or let them expire. I don’t let many go, although in my early years of domain name investing, I did let many more go.
I tend to focus on 5-8 letter domains (4 letter domains are almost all taken up, the worthwhile ones that is, but I don’t spend any time searching just for these), but I might go beyond 8 letters if the name sounds very good to me.
I do have some domains that are real words, including people’s names, but this is less than 5% of my portfolio.