What we measure

Six sets of criteria, drawn from naming practitioners and from trademark law. None of them is authoritative — they are working heuristics, and they contradict each other often enough that the disagreements are the interesting part. The one place we lean on evidence rather than opinion is the research.

The seven dimensions

The spine of the scorecard, because it is the only set that arrives with worked examples we can check ourselves against.

EVOC: Evocativity — Conveys at least a hint of what it's naming.; judgment
BREV: Brevity — Shorter = better.; computed
GREP: Greppability — Not a substring of common words.; computed
GOOG: Googlability — Reasonably unique (and domain name available).; judgment
PRON: Pronounceability — You can read it out loud when you see it.; judgment
SPEL: Spellability — You know how it's spelled when you hear it.; judgment
VERB: Verbability — The name (or variant thereof) can be used as a verb.; judgment

Their author ranks them too, but our weights don't follow his ranking, and his own ratings no longer set the standard. We keep the seven questions and judge the answers against a hand-rated calibration set; the weights and the reasons for them are on Methodology. Domain availability, which he folds into Googlability, is scored as its own factor.

A four-grade ladder, and the .com rule

Both a cool word and refers to what the company does (Writely)
Merely a cool word (Del.icio.us) — "This is enough… we're happy"
Mediocre but not actively repulsive — names "take on a tint of whatever they refer to"
Actively repulsive, sometimes on purpose (Webshaka, picked so they wouldn't keep it)

Plus a separate rule: own the exact-match .com. We measure that directly — see the gradient, and the case against it.

Seven brand criteria

Distinctiveness, brevity, appropriateness, easy spelling and pronunciation, likability, extendability, protectability. Note that brevity and spelling/pronunciation appear here and in the seven dimensions, arrived at independently — that convergence is worth more than either list alone. One claim here is testable: high-imagery names beat abstract ones, and imagery tracks etymology. We measure it with concreteness norms.

SMILE and SCRATCH

The only framework framed as deal-breakers, which makes it the most useful for tagging. SMILE: suggestive, memorable, imagery, legs, emotional. SCRATCH: spelling-challenged, copycat, restrictive, annoying, tame, curse-of-knowledge, hard-to-pronounce. "Spelling-challenged" says more on a scorecard than any number does.

Trademark distinctiveness

generic → descriptive → suggestive → arbitrary → fanciful, weakest to strongest. This is the legal ground under Neumeier's "protectability", and it is where the frameworks collide hardest: a descriptive name can score well on Evocativity and be almost impossible to defend.

Levels of engagement

Functional, invented, experiential, evocative — on a −2 to +5 scale. Functional names ("center", "medical", "health") sit at the bottom; evocative names that suggest a position rather than a function reach the top. A categorical axis rather than another quality score.

The case against the .com rule

The .com rule is the loudest claim on this site, and it was contradicted four years before it was made. Ev Williams — Blogger, Twitter, Medium — gave five reasons domains were already mattering less:

Google is the address bar. "Google is how normal people get around the web" — people search for a site rather than typing its domain.
Auto-complete. After a visit or two the browser remembers, so nobody has to recall the exact string.
Mobile hides the address bar, removing the constant reinforcement a domain used to get.
Apps. Discovery moved to app stores, where a domain is irrelevant.
Alternative TLDs demonstrably worked — bit.ly, del.icio.us, last.fm, turntable.fm.

Our own data cuts both ways. Exact-match .com ownership has collapsed from 94% of the 2006 cohort to under 25% today, which is either evidence that founders stopped caring for the reasons Williams gives, or evidence that the good names simply ran out. The measurement cannot tell those apart, and neither can we.

Rules worth ignoring

Naming advice accumulates folklore. Two of these would actively mislead anyone reading the scores here:

"Your name should return almost no Google results." What matters is owning the first page, not minimising the total. A rare string that someone else outranks is worth less than a common one you dominate. Our Googlability score measures collision, not scarcity — do not read it as the latter.
"Six letters maximum." Memorability tracks meaningful units, not character count. YouTube, Facebook and Wikipedia all break it. Our Brevity rule counts letters and syllables, and allows eight letters at two syllables.
"Start with an early letter of the alphabet." A phone-book-era rule about alphabetical listings, obsolete since search.
"Certain letters are lucky." There are no magic letters.
"A name should be an empty vessel." Meaningful names work through the associations they already carry.
"Whole categories of name are bad." Latin roots, blends and compounds all succeed when executed well.

And the case that none of it matters

The author of the seven dimensions ends his own article this way: "none of this really matters. You may have gleaned that from all the examples of smashingly successful names that ignore any number of these desiderata."

The stronger version of the argument is that it is what you do, not what you call it, and that being memorable matters more than what people remember.

Hold both. A name is worth an afternoon and not a month, and every score here is an opinion about a string, not a forecast about a company.