DomainValue
Menu

Guides

The doctrine, end to end

How this desk makes and judges names — every gate, register and method, rendered in full below.

Contents

The gates

Laws are the hard rules, above all mechanical gates. Tests run on every survivor.

Laws

0law

GREATNESS — never settle for generic

The hardest rule, above all others. NEVER settle for a generic, category-descriptive, sounds-professional name. Keystone / Meridian / Summit / Vantage / Vertex / Zenith / Atelier / Foundry / Cornerstone / any holding-Latin-mush = BANNED even when they clear every other gate — clearing the gates is the FLOOR, not the bar. The bar is GREAT: a name carrying a story, a POV, an unexpected pairing, a cultural-or-mythic reference that a client leans in and asks about, then repeats. Benchmarks: ByteDance · Bending Spoons.

0blaw

ACTIVITY, not location

The name must express the ACTIVITY — what the business DOES — not the location of incorporation or the founder's geography. ByteDance is Chinese but named for its activity, not "China". Never wear HK / China / Orient / Pearl / Jade / Cathay / Kowloon / Silk / Eastern / Monsoon labels just because that is where the paperwork sits. NAME THE VERB.

0claw

Real words, not elvish coinage — the adult-warmth test

Invented/coined sounds read elvish-fantasy-geek and land COLD on a real adult business person. Regno / Aurel / Novar / Auren / Embra / Valen / Velor / Solen / Ryven — and the parfum-Latin Noirael / Aurvex / Noctavo / Sovaria / Ictova batch — all failed here: they mean nothing, evoke no image. Warm names are made of REAL, RECOGNISABLE WORDS that fire a concrete image the moment they are heard (a lotus in bloom; bytes dancing).

0dlaw

Smart tech, not geek

When one half is a tech word, the WORD CHOICE decides geek vs mogul. Use broad, warm tech words an adult instantly pictures — pixel, signal, data, cloud, code, stream, pulse, wave, spark, current, digital, vector — NOT programmer jargon (kernel, hash, node, socket, cache, buffer, daemon, packet, root, commit, deploy).

0elaw

Laidback, not try-hard — the menace-tranquille test

Trying to sound powerful is the tell of the geeky kid who "wants to reign over the world". Flex words are BANNED as the soul-half: Reign, Crown, Throne, Empire, Dominion, Sovereign, Rule, Iron, Prime, Apex, Titan, Force, Power, Supreme, Elite, Alpha. The truly iconic names are laidback — ByteDance, Google, Amazon, Apple, Stripe, Meta, Uber — they never announce power, they imply it.

0flaw

The phone test — one word, writable on first hearing

The first filter run on every candidate. Say it ONCE on a phone; the listener must write it correctly with zero spelling help. ONE WORD ONLY (or a tight two-ultra-common-word compound). AUTO-FAIL: multi-word phrases · idiom-sentences · "The X of Y" · verb phrases · homophone-ambiguous letters (c/k/s, ph/f, -le/-el/-al, i/y) · silent letters · accents. A plain writable name beats a clever unwritable one every time.

0glaw

Not name slop — the anti-AI-averaging gate

AI namers do linguistic averaging: every output is short, punchy, vaguely technical, ending in a vowel or an x, or an [Adjective][Noun] compound (NovaTech / BrightPath / SwiftFlow). Documented-saturated 2026 tokens to AVOID: Nova · Flux · Pulse · Nexus/Nexum · Atlas · Orbit · Vectra/Vanta-class · Synth · Neural · Axera. THE TEST: would an off-the-shelf AI generator have produced this from a one-line brief? If yes = slop, kill it.

0hlaw

First read — the dominant sense wins

A bare brand name is a zero-context read, and the DOMINANT (most frequent everyday) sense of each word wins the race. billing → invoices (not theatre credits); market → finance (not the town square). The intended read must BE the buyer's dominant everyday read. Operationalised by the cold-read probe (zero-context, third-person, ≥3 model families).

0ilaw

Moldu friendliness

The name must read WARM and human to a non-tech regular person (a hairdresser, a shop owner — the actual buyer), not tech-SaaS-clever. Moldus do not buy workflow, they buy friendly help.

0jlaw

Heard spelling — one obvious typing

Extends 0f: the spoken form must map to exactly ONE obvious spelling; a silent/dropped-vowel trap = fail (mane heard as "man" → manflo). Check the organic typo variant too: if users will mistype it, the wrong form must not be a competitor's domain.

Tests

buyer_invoice_testtest

The buyer invoice test

The killer gate. Say it out loud to a REAL client — "you'll see a charge from ______". The name must sound like a credible BUSINESS SERVICE on a card statement, not a consumer product. (Belvolta = water bottle · Medaljet = printer · Titanwon = crypto casino · Drumnova = music startup.)

both_parts_earn_their_placetest

Both parts earn their place

The automano law: every syllable must MEAN something (auto+mano = car+hand; hs+mate = HS-code+helper ≈ checkmate). Filler suffixes (-ano, -ora, -ify, -ly bolted on) = instant fail.

radio_testtest

Radio test

Spellable on first hearing over the phone. Kills x/q clusters, Latin coinages, ambiguous c/s, obscure words.

esl_testtest

ESL test

Made of simple international words (A1/A2: top, gold, star, king, fame, bell, medal, drum) or a perfectly phonetic coinage. Kills hawk-class words ESL buyers cannot picture or spell.

french_read_testtest

French read test

Say it with a French ear. Kills Saymont (≈saumon/salmon), TopSaid ("said" reads the Arabic name Saïd), Salvolta ("sale" = dirty). Any comic or off homophone in French = dead.

letter_phonosemanticstest

Letter phonosemantics

The initial letter carries the trust register: Z opens edgy/gaming/crypto; soft-solid initials (B, M, V, S, F, L) open trust. Match the initial to the register.

length_and_formtest

Length and form

TLD per RANKED_TLD_POLICY: .com default > <name>qualifier.com > .ai for AI-native > .io dev-only > .co weak; blacklist .xyz/.org/.net/novelty as a brand. ≤8 LETTERS (soft — 9-10 if genuinely distinctive), letters only, one word. Compounds: BOTH parts ≥3 letters AND both real meaningful words; 4+4 balanced splits beat 3+5.

register_fittest

Register fit

Descriptive compounds read tool/SaaS (caps pricing power); pretty coinages read consumer/FMCG (fail the invoice test). Pick the register from the BUYER/brand-type BEFORE generating.

swap_testtest

Swap test

Replace the name with a competitor's in your messaging — if it still works, the name is generic.

wordplay_that_installs_itselftest

Wordplay that installs itself

The hsmate winner-factor: a play the listener gets free (hsmate ≈ checkmate) = zero marketing budget to explain. ⚠️ Polarity condition: the play only counts if gate 0h passes — both senses positive AND the dominant everyday sense is the intended one.

fluency_kill_ruletest

Fluency kill rule

If a stranger cannot pronounce it correctly on first sight of the written word, kill it. Alter & Oppenheimer: fluently-named stocks measurably outperformed disfluent ones post-IPO. No "it's pronounced…" brands, no misspelled real words.

context_render_testtest

Context render test

A name that survives list review can die instantly IN SITU — render finalists on a mock page/logo/invoice before locking. Cheapest gate with the highest observed save.

rankability_ai_entity_testtest

Rankability / AI entity test

Can you OWN the name's SERP and be entity-unambiguous to LLMs? Google it AND ask 2-3 LLMs "what is <name>?" pre-commit. Dictionary/common words (Apex, Nova, Orbit) are high-collision: the model resolves to a clearer, better-known entity.

outcome_not_mechanismtest

Outcome, not mechanism

Name the OUTCOME/promise, not the mechanism. Bonus: narrative scaling — prefer a name that auto-writes future messaging.

sound_symbolism_palettetest

Sound-symbolism palette (generation input)

Match mouth-feel to positioning: front vowels i/e = small/fast/light/sharp; back vowels o/u/a = big/rich/smooth/powerful. Plosives p/b/t/d/k/g = impact; fricatives f/v/s/z = speed/elegance. v = alive, b = reliable, z = attention, x = innovation.

position_against_the_categorytest

Position against the category

Map competitors on the taxonomy Functional → Invented → Experiential → Evocative and name where the category ISN'T. Lexicon's polarisation signal: a name everyone is instantly comfortable with is wallpaper. NEVER focus-group a name.

linguistic_screen_all_marketstest

Linguistic screen — all markets

Native-speaker check in EVERY launch language. Mandarin + Cantonese are MANDATORY on any Latin name, and Cantonese is checked SEPARATELY from Mandarin. The disaster mode in Chinese is almost always homophonic, not semantic.

The three registers

The generative frame — pick the register before generating a single candidate.

A

Soft romance coinage

Invented, SOFT consonants (m·r·t·l·n·v — "premium feel, no harshness"), vowel-ending, 5-9 letters, brand-blank or a root with mythic depth.

Exemplars

mirtela (myrtle→Aphrodite + tela/canvas) · furnely (fur + lovely/homely) · mediferra · susabe · funuli · lulabu

For

Consumer / DTC / premium / fintech-blank.

Risk

Reads consumer/FMCG → fails the B2B invoice test.

B

Self-explaining compound

Two REAL parts that both earn their place, product category BAKED IN, wordplay that installs its own positioning.

Exemplars

automano (auto+mano = automation+hand) · hsmate (HS+mate ≈ checkmate) · lexroi (lex+roi = law king)

For

Tools / products.

Risk

The both-parts law is mandatory here.

C

Concrete greatness / myth

REAL, SIMPLE ENGLISH words (ESL-instant) that NAME THE ACTIVITY with a clever, unexpected, memorable twist — short, punchy, authoritative.

Exemplars

SpaceX = space + X (eXploration AND a frontier symbol) · ByteDance = byte + dance (the whole activity, in two ESL words)

For

Holding / agency / ambitious B2B house.

Risk

Generic-professional (Thrive/Victorious/Meridian) is the bland lane — rejected.

The construction doctrine & methods

How candidates are produced, in full.

primary

Distant-leap blend

The #1 method for a free + great + .com name in 2026. The exemplars are BLENDS of two real words, NOT single dictionary words: ByteDance = byte(tech) + dance(art/social); Pinterest = pin + interest; Instagram = instant + telegram; Airbnb = air + bnb. Single real words are all taken; the novel BLEND is free precisely because nobody paired those two words. THE KEY = DISTANT LEAP: the two words must come from FAR-APART domains that resolve into the business (tech↔dance, not trade↔wind). Obvious same-field pairs are taken; the SURPRISE of the leap is what keeps the .com free AND makes it click.

Construction engine (pipeline v3)

THE CONSTRUCTION DOCTRINE. Every prior cycle mined one of two quadrants: (1) REAL WORDS (high density, all taken — 24/25 on .ai) or (2) LOW-DENSITY COINAGES (free but mush: tidezo/dawnva = zero phonological neighbourhood). The unmined quadrant: LEXICALLY EMPTY + PHONOLOGICALLY DENSE — constructed words that FEEL like real words (planity ~ plan/sanity/plain; glissando around lissando) while being legally empty. Mechanism: morpheme matrix → hard filters → score = char-bigram log-prob + log(neighbourhood density). Round 13 measured 18/26 FREE on .ai here — the domain wall is gone in this quadrant.

Racehorse winner method

For house/holding/flagship names: generate like naming a RACEHORSE — a destiny-claim to be SHOUTED by the announcer at the finish (Secretariat! Man o' War! Empire Maker!). Recipe: (a) an APEX wordlist (money/space/empire/king/dark ambition words) × (b) a PERCUSSION wordlist (shot/X/dance/horse/strike punchy action) with the pair INTERACTING (verb grammar, never noun-concatenation); (c) ANNOUNCER-TEST each survivor aloud ("AND HERE COMES ___!"); (d) tri-civilisation echo check; (e) collision check.

Volume rule

Generate 1,000+ candidates, shortlist 50-100. A 20-name brainstorm is structurally under-powered. YC data: 87% choose brandable over keyword; median 8 chars (6-10 sweet spot); 2026 momentum = unexpected compound pairings (specific enough to feel real, broad enough to pivot). Compounds are "1+1=3" — richer associations beat shorter-but-empty coinages.

Three-team method (anti-anchoring)

The documented fix for BOTH human groupthink AND AI linguistic-averaging is identical: DISGUISE / DIVERGE THE BRIEF. Run parallel lanes: one with the real brief, one "naming a competitor", one naming an UNRELATED category (Windsurf came from the unrelated team).

Cold-read probe (0h, operationalised)

Run on every finalist BEFORE the collision sweep — cheaper and kills faster. Protocol: (1) zero-context probes, one question per FRESH session, ≥3 model FAMILIES (not 3 runs of one), ≥5 runs each at default temperature (the SPREAD is the signal); (2) third-person framing always — "a company is called X: what do they sell? what industry?" — NEVER "I am considering naming my company X" (sycophancy caves the test); (3) the shortlist probe: does the name even belong on that register?; (4) score by dominance of the first guess.

TLD policy

SUPERSEDES the old ".com-only forever" rule (relaxed 2026-08-27): .com fell to ~46% of YC startups and Placek calls .com just an area code. The NAME-quality bar (gate 0) is unchanged — this governs ONLY the TLD once a great name exists, and it is BUYER-MATCHED.

  1. 1..comThe DEFAULT for any brand — plurality, highest type-in trust.
  2. 2.<name>qualifier.comStrong second (get<name>.com / <name>hq.com).
  3. 3..aiFor an AI-native brand — serious AI companies ship and stay on .ai.
  4. 4..ioDev-only.
  5. 5..coWeak.

Blacklist: .xyz · .org · .net · novelty TLDs as a brand

Full evidence and the ranked table live on the market data page.