For each name you want to try, follow these guidelines:
1) say it out loud to ten people and ask them to spell it
2) keep talking to them for about a minute, then ask them if they remember it
3) write it down for ten people and ask them to pronounce it
If you get less than 90% right on #1 or 3, your name will get lost in translation and people won't be able to find it even if they want to (or, you'll have to make a lot more impressions before people start remembering you). If you get bad results (probably won't be as high because of peoples' memories) on #2, your name it probably to long to remember or not differentiated enough.
Rules to Remember:
1) Easy to spell correctly if heard
2) Easy to say if read
3) short = easy
4) real words = easy (but usually conflicts with #3 because of availability)
5) Don't be blatantly about your business, but related and give hints (Xobni and Auctomatic are good examples of doing this right - much better than emailsocialnetwork.com or auctiontools.com)
6) Getting a domain name that doesn't follow these should be considered a marketing expense, because you'll have to make more impressions in order to get planted in people's consciousness.
Here's my take on some YC company names
Loopt - probably lost some people who thought it was looped, but only 5 letters and one syllable so it's easy
Weebly - weird word but hard to get wrong, good name
Xobni - like Loopt, probably lost some people looking for Zobny, Zobni, but the "inbox" thing makes it easy to remember
Wufoo - probably the hardest one (on this list) to get right, but being short is a virtue
Auctomatic, octopart - longer names that are made of real words are easier to remember and spell
1. picked a list of 50-60 words/prefixes/suffixes we liked
2. wrote a python script to generate permutations, and then
3. ran dig -f on the text file to see which weren't taken
4. scraped buydomains search results to see which ones were easily for sale.
absolutely everything else was a waste of time, although we're also working with a naming agency and that might be promising (but expensive).
amusingly, domains that were occupied by squatters or companies like buydomains were easier to deal with -- at least they're rational and the conversation, though annoying, is straightforward (arrive at a #.)
domains that are occupied but unused (and don't have a 'for sale' sign hanging up) are generally hopeless. if you get a response, you'll generally get some whacko who either won't sell it, or who will only sell for millions of dollars or something. you'll probably not believe me, and will learn this the hard way (just as pg told us months ago :))
hivgene.com was available and directly related to the first iteration of our startup.
More recently, we were looking for names for something of an expanded version of the startup and my cofounder and we sat plugging away at instantdomainsearch.com and IMing each other available names as we found them. This got us xygene.com. I couldn't have asked for anything better.
I think we're just really, really lucky with domain names.
I suggest you run a randomized test using Google keyword ads. Try different names taking you to some page and see which ones get the highest click through. Ian Ayres used this technique to find the title for his last book. Seems like a cheap and effective way to do it to me.
speaking as someone who tried this route, don't bother. if you have a product named xyz, people expect it to be at xyz.com.
otherwise, you will lose the traffic that assumes xyz.com is your domain.
and not just from random visitors but important contacts like investors, who assumed we were at dropbox.com. even people who knew well that we were at getdropbox.com would routinely fire off emails to dropbox.com by accident.
the process sucks, but we're changing our name after wasting a lot of time hoping we could secure dropbox.com and going back and forth with the owner.
Has anyone tried taking an analytical approach to finding a name? This is an approach I am currently trying - registering a lot of domain names and then using Google adwords to find out what name most people click on. Have no idea how well it works yet, but it seems much better than relying on your 'gut' to choose a name. Here's the approach:
1. Find some available domain names that are, at the very least, easy to pronounce and spell. Easier said than done of course, but I still think it's possible to find some good stuff. For example, this website - http://www.5letter.com/domain-names/ - allows you to search through all available domain names that are exactly 5 letters long. There are some surprises there.
2. Register as many domain names as budget allows. If it sounds good and its available, grab it.
3. Throw together a website with a bit of info about your product and a box for people to enter their email address to find out when you're launching. I'm sure you have seen these kinds of sites before. The whole point of doing this is to make sure people have something to look at when they click your ads, and you might get to collect a few email addresses to notify when you DO actually launch.
4. Go to Google adwords and set up an account (http://adwords.google.com). Create an ad that explains something about your product. Clone this ad for each of your domain names - ensuring that the ad title contains ONLY the text in each of names to test. For example, if your domain name is XYZ.com then put XYZ in the title. The point of this is that all your ads will be identical EXCEPT for the title, which contains a potential product name.
Make sure all the ads point to the website created at step 3. Choose keywords that people in your target market will be most likely to search. Make sure your budget is high enough to ensure a reasonable statistical sample for each ad. You need enough, in my opinion, to collect at least 100 clicks per ad. Google can handle all of the ad rotation for you, though there's no guarantee that impressions will be equal for all ads.
5. (I'm currently at this step). Launch all of the ads. Wait for a couple of weeks or for however it long it takes to get a statistically reasonable number of clicks and impressions. Once you're happy with the numbers, look at the click/impression ratio for each ad. Basically, the ad with the highest ratio of clicks to impressions "wins", and the name in the ad title is your product name. If there's no clear cut winner, the idea is to cut all the losers and re-run the top 2 (or 3 or whatever it is) until a clear winner is found.
A couple of things about this method:
a. The method costs a bit and can take up a bit of time. I've spent about $300 so far, and some of the ads still haven't got enough impressions yet after two weeks.
b. You obviously don't need to actually register all of the domain names, but I still think it's a good idea just in case you miss out on a good one. You don't need to create the website either, but I think it's a good idea if you're serious.
c. I'd imagine this process might work well for product names in general.
d. There's a chance you'll get to a sort of local maxima with your choice of names. No idea how to resolve this, short of coming up with as many different ideas as possible.
e. Probably won't work too well for products aimed at hardcore geeks, as they don't tend to click ads or block them altogether.
Yeah - so far there are about 3 clear leaders (30-40% higher than the others). I probably need to run the ads for another two weeks before I know for sure, though.
you are not considering your position. if you are number one for your keyword your click through ratio will be higher no matter what. assuming all else is equal, bids, keywords, ads. your average position will be higher or lower depending on your quality score which is determined by several opaque factors which are hard to quantify.
i just dont know if adwords is the best way to test brand affinity.
2) keep talking to them for about a minute, then ask them if they remember it
3) write it down for ten people and ask them to pronounce it
If you get less than 90% right on #1 or 3, your name will get lost in translation and people won't be able to find it even if they want to (or, you'll have to make a lot more impressions before people start remembering you). If you get bad results (probably won't be as high because of peoples' memories) on #2, your name it probably to long to remember or not differentiated enough.
Rules to Remember:
1) Easy to spell correctly if heard
2) Easy to say if read
3) short = easy
4) real words = easy (but usually conflicts with #3 because of availability)
5) Don't be blatantly about your business, but related and give hints (Xobni and Auctomatic are good examples of doing this right - much better than emailsocialnetwork.com or auctiontools.com)
6) Getting a domain name that doesn't follow these should be considered a marketing expense, because you'll have to make more impressions in order to get planted in people's consciousness.
Here's my take on some YC company names
Loopt - probably lost some people who thought it was looped, but only 5 letters and one syllable so it's easy
Weebly - weird word but hard to get wrong, good name
Xobni - like Loopt, probably lost some people looking for Zobny, Zobni, but the "inbox" thing makes it easy to remember
Wufoo - probably the hardest one (on this list) to get right, but being short is a virtue
Auctomatic, octopart - longer names that are made of real words are easier to remember and spell