Skip to content

Methodology

Every rating on this site is an opinion about a name. Nothing here is a statement of fact about a company, its product, its people, or its prospects, and no score should be read as a prediction about any business.

We score the name and only the name. A name is not good because the company succeeded, or bad because it failed.

The basis

The seven questions come from Daniel Reeves' "Nominology": is the name evocative, brief, greppable, googlable, pronounceable, spellable, and could it become a verb? To those we add an eighth factor, exact-match .com ownership, from Paul Graham.

Reeves supplies the questions, not the answers. What counts as green, yellow or red is set by a calibration set of 100 companies rated by hand by a frontier model, with a written reason for each rating, so the standard can be read and argued with rather than taken on trust.

How each dimension is rated

Evocativity
Judged
Does it suggest what the company does? Rated by a language model given the definitions and the calibration set as examples, with one short reason per rating.
Brevity
Computed
Letters, not characters: up to seven is green, eight is green only at two syllables, up to ten is yellow, eleven or more is red.
Greppability
Computed
How many words in Webster's Second (1934) contain the name, and whether it is a word itself. The dictionary's age is the point: it predates every modern brand.
Googlability
Judged
Is it free of a prior meaning competing for it in search? Rated by a language model given the definitions and the calibration set as examples, with one short reason per rating.
Pronounceability
Judged
Can you say it on first sight? Rated by a language model given the definitions and the calibration set as examples, with one short reason per rating.
Spellability
Judged
Can you spell it after only hearing it? Rated by a language model given the definitions and the calibration set as examples, with one short reason per rating.
Verbability
Judged
Could it become a verb for the product? Rated by a language model given the definitions and the calibration set as examples, with one short reason per rating.

Googlability used to be computed as membership of the same 1934 dictionary. Against the calibration set that missed surnames, places, initialisms and every word newer than the dictionary, so it is now judged: whether a name collides with a prior meaning is world knowledge, and a model has it.

The two scores

Nominology combines the seven dimensions and .com ownership, out of 100. Green earns a factor's full weight, yellow half, red none. A dimension not yet rated, or a .com we couldn't check, is left out rather than guessed, and the card says it is partial.

FactorWeightWhy
Pronounceability 20 With brevity, one of the two properties with published outcome evidence behind them.
Brevity 15 The same fluency evidence: short names are easier to process and to remember.
Spellability 15 A name heard but misspelled loses word-of-mouth traffic.
Googlability 15 A prior meaning competes with the company for every search.
Exact .com 15 Owning the exact .com earns trust and typed-in traffic (Graham's 2015 rule).
Evocativity 10 Helps at launch; over time a name takes on the tint of whatever it names.
Greppability 5 Overlaps Googlability, and matters mostly for text search.
Verbability 5 Red for most names; it rewards success more than it predicts it.

Fluency uses brevity and pronounceability only, the two properties with published outcome evidence behind them (see Research). It is kept separate so the other factors don't borrow an authority they haven't earned.

Exact .com

Measured as Graham states it: does the company's own site sit on the exact-match .com? We read that off the company's website. For a name typed into Check a name, we ask the .com registry whether the domain is registered at all. It is its own factor rather than part of Googlability, so a name and its domain can be judged separately.

Limitations

Archive: Reeves' own ratings

The site first calibrated against the ratings Reeves published for 19 names. They were one practitioner's judgment, and unevenly generous: the only all-green rows are his own TagTime and Graham's Viaweb. They no longer decide anything. They are kept here beside the same names re-rated by hand, so the change can be checked.

Compared on the four dimensions both sets judged: they agree on 61 of 76. An outlined tile marks a disagreement.

In the Opus row, Brevity and Greppability are today's computed values and Googlability is blank: those names were re-rated on the four dimensions both sets judged at the time.

Where the criteria come from

The seven questions are Daniel Reeves'; the grade ladder and the .com rule are Paul Graham's. Both are essays, not evidence. We use them because they are specific enough to test, not because they are authoritative. The published findings we cite are on Research, and the company lists on Sources.