1) Choose a name that reflects both your brand soul and what your business actually does. “Brand soul” is the spirit you want your company to have: energetic, happy, meditative, efficient, etc. This is hard to quantify but is made up largely of a) associated meaning and b) sound. “Metaphorical” names can fit this bill. A metaphorical name is a real word or compound of words that relates to your brand and to what your brand wants to say- but is not purely literal.
Java is a great metaphorical name for a product – fun to say, easy to spell, and while it’s both a little bit exotic and slang, everyone knows that it means something that wakes you up- something exciting.
Basecamp is another great product name in this category. It connotes structure and reliability, yet a sense of adventure. A basecamp is what you return to for comfort after, say, a foray up Everest. A project/task management tool is what keeps you sane while climbing the young-company mountain. (Note that Basecamp doesn’t have basecamp.com – they use basecamphq.com – and still returns at top of search.)
Personally, we’re not haters of all “nonsense” names, because they can be playful and fun-sounding. I think the much-maligned Google is a great name: it’s fun to say, it can be used as a verb, it has a meaning that’s cool for early adopters and employees, and they get to call their campus the Googleplex.
Want some terrible nonsense names? Flisp (makes you feel dyslexic just reading it). Plurn (playlist + burn). Oooooc.com.
Don’t go for a soulless corporate compound name (this seems to happen most often in rebrands): Accenture, Brand-Image, Integy, Xfinity. You want love for your brand. It has to have a personality.

2) Try to stay away from naming trends, particularly ones that arose from a dearth of available domains (i.e., the post-Flickr dropped-vowel trend). Not only do they get annoying, it’s instantly clear what year your company was founded.

3) Avoid the “Chevy Nova incident”**: Run your proposed name through as many translators, human and digital, as possible.

4) Pronounceability: If you want a global brand, think about people who do not natively speak your language (whatever that might be). Avoid sounds that are hard to say in your target markets. We had this issue with HauteLook, but the company gained so much traction that management decided against a name change.

5) Spelling Bee: Likewise, if you say the name to someone and they later search it, do they have a hope in hell of spelling it the way you do? Three-and four-letter names get a bit of a pass on this.

6) Uniqueness: Consider how your user will search for you. If you’re consumer Internet, for instance, it’s very likely they will type the brand name into the search bar rather than search for “cool Net utility I might like”. While we’re all for a name that relates to your business, don’t use generic words. (If we called ourselves “Strategy”, we’d be unsearchable. If you search for str.ate.gy, though, you will find us – although if we were the kind of company that gets business through search, we wouldn’t have named ourselves that, nor used a domain hack. We are actually called this because we didn’t want to fall into the agency trap of names like 27GreenFrogs and CandleMoo***. )

7) The “Beavis & Butthead” filter: If it would make a ten-year-old boy snicker, that’s not good.* You can be immature enough, just this once, to do this test. If you can’t, a) you might want to get help with that; b) go find someone with a refreshingly juvenile sense of humor and read your name shortlist to them. Names that fell into this pit: Fairtilizer, Tailrank, Ainol, Flowtown, and Nook*.

*Yes, a lot of people think iPad falls into this category. Apple isn’t perfect.

**The Chevy Nova story is actually an urban branding legend. The Mitsubishi Pajero, however, is a cautionary tale. As is WaterPik in Denmark.
***These are made-up names. I’m actually kind of liking CandleMoo.

We do quite a bit of naming here- companies and products- but we don’t usually blog, tweet, or generally gab that much about the process. Having hired a few naming firms myself (or been subjected to someone else’s hire), I’ve learned a lot about what works, what doesn’t work, and what is simply wasting time.
A few key things to remember in choosing a naming firm:

1) Work with someone who sets clear deliverables. Know how many names you are going to see on first pass, second, etc., what you need to do in response, and what the process is of getting to a shortlist.

2) Choose a firm that can work to your understanding of what a name should be. Given free rein, we tend to go for names which are both cool words and reflect what exactly your business does. Yes, you can brand names which have zero to do with your business, and that can be great, but if you don’t have to, why would you? When we named WeFi, a wifi hotspot-based social network, we got it all- a name that reflected the biz, a cool word that was fun to say and easy to spell, and a FOUR LETTER DOMAIN which we were able to purchase for about $2500.00.
That said, we have a lot of fun creating names that are less literal, have hidden meanings, or are just real imagination-catchers. You want a company that understands your parameters and priorities, and the kind of names to which you gravitate; a company that works with those guidelines- without delivering bad names just to please you. Also, if you’re going to purchase a domain to go with the name (as you almost certainly are), clarify whether you must have the .com (ideal of course) or whether you are open to a .net, a version of your name (getnewco.com), or even a domain hack (like our own http://str.ate.gy). We don’t love domain hacks except for very specific, somewhat renegade businesses, but it worked for del.icio.us (who eventually were purchased by Yahoo! and could then quite easily afford to buy delicious.com), and it might work for you. Make sure your namers know your stance on this.

3) Choose someone who understands and respects your resources and time.
For instance, regarding domains, you shouldn’t be given a bunch of names far outside your budget or (pet peeve here) names that have not been vetted at all for availability. This particularly applies to “parked” domains, because many of the early mass domain acquirers do not sell even when offered a huge multiple of what they might get from search. They think they have oceanfront property and you will never convince them otherwise. Once you are into the home stretch and looking at (in our process) 20-40 names, you should have at least a ballpark idea of whether each domain can be purchased and what it will cost (or if miracle of miracles, it’s available). Throw your namer a hundred bucks or so to register anything that’s actually available that’s close to the shortlist. We usually register all the domains we think are really cool ourselves for later use, with a clear clause in our contract that we have to GIVE the name our client chooses to the client.
There may be a few names that you love where the domain holder can’t be reached easily or is a pain to negotiate with, and in our naming projects we will include these, because and only because they are great- but we are always upfront that this particular name may just not be possible and that there must be a backup. We never want a client to fall in love with a name only to be told, with no prior awareness of issues, that it’s just not going to happen, or that the price tag is far out of their budget.

4) Once you are into the “homestretch” of 20-40 names, someone on one side or the other should have done at least a cursory trademark search. This can be done via the U.S. Patents & Trademarks Office (USPTO) and some judicious search. Clarify with your namers whether this is your responsibility or theirs. Trademark is a complicated area and can’t be assessed rapidly, but, like the unavailable or out of reach domains, why waste your time on a name when there is a clear trademark issue one search away?

5) Finally, it should not need to be said, but work with a company that follows the most basic rules of naming, which will be in the next post or so, but which include things like avoiding the “Chevy Nova incident” and running the name through the “Beavis & Butthead”/ten-year-old boy filter.

Happy naming!