0.74
Pearson r · score ↔ aftermarket resale price
A 10-point score lift correlates with a 38% median resale premium across 41,200 transactions. Stronger than category, TLD, or word length as a sole predictor.
Methodology · v3.2 · Updated March 2024
The Brandability Score is a calibrated instrument — not a feeling.
A 0–100 composite that grades a domain on phonetics, memorability, trademark risk, and SEO headroom — built on 41,200 aftermarket transactions and the same heuristics our team used at GoDaddy and Landor before we left to ship something better.
The four inputs
Four independent measurements, one defensible number.
Each pillar is scored 0–100 on its own scale, then weighted. No single heuristic decides the outcome — a name that shines on phonetics but trips trademark still gets dinged. That's the point.
-
01
Phonetics
Weight · 30%How the name sounds when spoken aloud. We measure syllable count, consonant-vowel balance, hard/soft consonant clusters, and whether the stress falls on an open or closed vowel. Names like
Stripe,Figma, andVercelscore 90+; names that require a pronunciation guide score under 50. -
02
Memorability
Weight · 30%How likely a stranger is to recall the name after one exposure. Tested via delayed-recall studies with 1,200 participants across English, Spanish, and Mandarin cohorts. Shorter names with distinct consonant clusters outperform longer ones — but only when they're also pronounceable.
-
03
Trademark risk
Weight · 25%The probability that the name collides with an existing USPTO, EUIPO, or WIPO registered mark in a related class. Higher scores mean lower risk. We index 4.1M active marks and refresh weekly. Note: this is a similarity heuristic, not a legal opinion.
-
04
SEO headroom
Weight · 15%The realistic ceiling on organic search performance for a brand built on the name. We look at exact-match search volume, exact-match domain authority precedent (for the TLD), and how generic-vs-proprietary the word is. A name like
car.comhas infinite volume and zero headroom;latimer.iohas none and total headroom.
A name's score should predict what a buyer does next — not flatter the founder who chose it. We don't grade taste. We grade outcomes, calibrated against 41,200 real transactions where somebody actually wrote a check.
Calibration
Calibrated against 41,200 real aftermarket transactions — not vibes.
The score isn't a heuristic we like; it's a model fit to outcomes. Every quarter we re-fit the weights against the previous quarter's Sedo, Afternic, and Dan.com sales. These are the headline numbers from the Q1 2024 audit.
14pts
Median score gap · scored vs unscored picks
Founders who filtered by score ≥ 75 ended up with names that scored 14 points higher than the unfiltered cohort — and 2.3× more likely to still be in use 18 months later.
50,000
Names sampled for distribution audit
A clean normal-ish distribution: mean 58, median 61, σ 16.2. We deliberately avoid a J-curve, which would mean the score only separates the top 10% from everyone else.
99.98%
API uptime · score endpoint · 36 months
The same availability database that powers our checker powers the score engine. Same SLA, same 6-minute refresh cadence, same public status page.
Methodology peer-reviewed by Dr. K. Andersen (linguistics, UT Austin) and the IP practice at Haynes Boone. Full audit log available on request to qualified researchers.
Worked examples
Familiar names, real scores.
Three names you already know, scored with the public engine. A household brand, a recent unicorn, and a name that scored well but flopped — so you can audit the model against your own intuition.
Household brand
stripe.com
94
/100 · Excellent
One syllable, hard consonant, owned-then-defined meaning. This is what the top of the scale looks like — and why trademark risk is the only thing keeping it from a perfect 100.
Recent unicorn
linear.app
82
/100 · Strong
Three syllables, dictionary word, generic meaning — the SEO headroom penalty is the model's honest signal here. Memorability rescues it.
Scored well, flopped
kuvio.ai
79
/100 · Strong
Phonetically beautiful, trademark-safe, pronounceable in five languages. Memorability dragged the average down — nobody could spell it after hearing it once. Shut down 2023. The score caught this.
Scores are post-hoc recomputations of public names against the current model. The original founders did not have access to the score at decision time.
What this score does not do
Limits, stated plainly.
A scoring model is only as honest as its stated blind spots. Here is what the Brandability Score intentionally does not measure — so you can decide when to trust it and when to bring in a specialist.
It is not a legal opinion.
The trademark pillar scores similarity against registered marks using phonetic and orthographic matching. It does not constitute legal advice, an opinion of registrability, or a freedom-to-operate clearance. If the name scores 95 on trademark risk, talk to an IP attorney before you print business cards.
It does not predict resale returns.
The calibration against 41,200 aftermarket transactions is a population-level signal — it tells you the median premium for a 10-point lift. It does not, and cannot, guarantee that any specific name will resell for any specific amount. Treat domain investment as a hobby, not a thesis.
It does not see live market depth.
The score measures the name. It does not measure who is currently bidding on the name, how many similar names are expiring this week, or whether the TLD is about to undergo a registry price hike. For those signals, use the drop monitor and the registrar liquidity panel — they're a separate product.
It does not replace your taste.
If you love a name that scores 62, name your company that. The score is a guardrail against costly mistakes (trademark collisions, unspellable coinages, names that compete with 800 incumbent brands on page one), not a creative director. We've watched plenty of 70-scored names outperform 90-scored ones when the founder actually cared about them.
Ready to run the numbers?
Score up to 25 names free, forever.
The free tier includes the full score breakdown, 350+ TLD availability, and the drop monitor. No credit card.