Inside the BetterNamer score
What the number next to each name means, which eight structural measures go into it, how the best available domain and the conflict checks change it, and why it is not a guarantee.
Every result in a BetterNamer search carries a score from 0 to 10. It is not an opinion from a language model and it does not move when the model is swapped. It is arithmetic over measurable properties of the name, the domains it can be registered under, and, on the paid tier, the conflict checks. This post walks through the arithmetic so the number means something when you see it.
Four inputs
The score combines up to four inputs, each on a 0 to 1 scale, each with a weight:
- Name quality. Eight structural measures of the name itself, described below.
- Domain quality. The quality of the best domain the name can actually be registered under.
- Web conflict. What a live search for the exact name found.
- Trademark conflict. What a search of USPTO records found.
On the free tier the score is name quality and domain quality in equal halves; the conflict checks do not run, so they contribute nothing. On Pro, the conflict checks each carry about a fifth of the weight and the two quality inputs drop to a quarter each. The weighted sum is multiplied by ten, then adjusted for the extension (below), then rounded to one decimal.
Name quality: eight measures
Each measure reads the letters of the name, nothing else, and returns a value between 0 and 1. Their weights sum to 1.
| Measure | Weight | What it rewards |
|---|---|---|
| Meaningfulness | 0.16 | One or two real words with high letter coverage (ShopFlow). Zero for a purely coined string. |
| Readability | 0.14 | Vowel and consonant balance; no awkward letter runs. |
| Pronounceability | 0.14 | Healthy alternation of consonants and vowels across the whole word. |
| Length | 0.12 | Four to twelve letters, peak at eight, steep falloff past twelve. |
| Memorability | 0.12 | Word-like shape, alliteration, recognisable affixes, a chantable syllable count. |
| Brandability | 0.12 | Punchy length, clean characters, distinctiveness, no consonant pile-ups. Independent of meaning. |
| Spellability | 0.12 | The radio test: can a listener type the name after hearing it once. |
| Letter preference | 0.08 | Common, easy letters over Q, X, Z and visually confusable pairs. |
Three design decisions are worth knowing. The measures are built not to double-count: memorability ignores whether a name is a dictionary word, because meaningfulness already scores that, and brandability ignores raw pronounceability for the same reason. The readability measure caps its bonus for a clean opening sound, because uncapped it saturated and let any name starting with a hard consonant and a vowel outrank better names. And spellability's largest penalty targets one specific pattern: real words welded to a non-word fragment (FinanSysAdvisors), which is the pattern that fails the radio test hardest. That measure was added after exactly such a name scored 9.0.
Domain quality: the best domain you can have
A name is only useful with a domain, so the second input scores the domains. Each candidate domain gets its own quality score (similar measures to the name, plus the extension, SEO friendliness and typing ease), and the name takes the score of its best available domain. A name whose strong extensions are all taken does not get credit for them; it is scored on what can actually be registered. Early in a search, before availability has come back, the best domain by score stands in, and the number settles as checks complete.
The extension nudge
After the weighted sum, the score is adjusted a little for the best available extension. Each extension has a preference value (.com highest, then .net, .ai, .org, .io, .dev, .co and so on, with a neutral default for the rest) and the adjustment is a few percent either way: a name whose .com is available is lifted about two percent; a name whose only available extensions are weak is trimmed slightly. The nudge is small on purpose. The name matters more than the suffix, and the base weights stay normalised.
Conflicts on Pro
The web and trademark checks each produce a conflict level: none, low, medium, high. Each level maps to a value: none is 1, low is 0.8, medium is 0.4, high is 0. A name with a high trademark conflict therefore loses that entire fifth of its score, and a name with a medium web conflict loses most of its web fifth.
The web check searches the exact name as a quoted phrase and grades a hit in a result title or hostname as high, a mention in a snippet or a near-match form (Summit Roast or Summit-Roast for SummitRoast) as medium, and nothing as none. The trademark check grades the closest match by how similar the name is crossed with how close the class is, and the conflict level is the worst match found.
A result the scoring does not yet express well, and which is on our list: on Pro, a name with no conflicts at all gets a flat credit for the conflict inputs, so a structurally mediocre name that happens to be clear starts from a higher floor than it deserves. The better design treats conflicts as a gate that can only lower a score rather than as credit that raises it, and that change is planned.
What 9.0 means, and what it does not
A 9.0 says: this name is readable, sayable, spellable, a sensible length, buildable from words or clean enough to coin, has a good domain available, and (on Pro) came back clear of obvious conflicts. It does not say the name is good for your business, that the brand will work, or that a trademark attorney would clear it. The score ranks candidates within a search so that you read the top of the list first; it is not a verdict on any one of them.
Two habits make the number more useful. Read the tabs, not only the total: a name can reach 7.5 with a great domain and a weak name, or the reverse, and the domain, trademark and web tabs on each result show which one you are looking at. And treat a medium or high conflict as the headline regardless of the total, because a 9.0 with a live trademark in your class is a 9.0 that belongs to someone else.
Tuning
The weights live in configuration, not in code, and they have changed before: the spellability measure and the steeper length falloff both arrived in August 2026 in response to names that scored well and should not have. If a name scores in a way that looks wrong to you, send us the name and the score through the contact page; those reports are how the measures get better.