Naming across borders: the checks that save you later

Every naming disaster story you've heard at a conference — the car that meant "doesn't go" in Spanish, the beverage that meant something anatomical in German — is at least half folklore. The Chevrolet Nova sold fine in Spanish-speaking markets; Spanish speakers were entirely capable of hearing the difference between "Nova" and "no va". But the fact that these stories are mostly myths shouldn't comfort you. It should worry you. The real failures weren't dramatic mistranslations caught too late. They were boring, preventable, administrative deaths: the trademark that was taken in class 9, the domain that belonged to an enthusiastic squatter, the pronunciation that turned every sales call into a spelling bee.

The wedding of naming practice with international reality is where our full naming process gets serious teeth. Here's the checklist we'd run before any name with ambitions crosses a border — much of it cheap, most of it doable in a fortnight, all of it cheaper than the alternative.

Screen the languages before you fall in love

The linguistic screen has tiers, and budget shapes which you run, but the floor is always the same:

Tier 1 — the desk screen (every name, every project). Machine translation against the top ten languages of your target markets, plus a slang dictionary sweep in your home market and the next three you expect to enter. Then — the step most teams skip — run the name through a search engine in those languages with incognito settings. What autocompletes is what your customers will find.

Tier 2 — native-speaker panel (shortlist only). Three to five native speakers per target market, a modest honorarium, and a standard questionnaire: What does this sound like? Is it difficult to say? Spell it after hearing it. Any unfortunate associations you're too polite to mention in business settings? Recruit through professional networks or linguistic screening services, not the one teammate who "did a semester abroad".

Tier 3 — dialect and register check (finalists only). The formal-register name that means "ledger" can be playground slang in a different region. Australian English gives us a built-in humility here; we've all watched a very normal local word travel somewhere else and become a scandal. Ask about rhyme risk too: what rude word is phonetically adjacent? Schoolyard cruelty scales into brand risk with remarkable fidelity.

The output isn't a pass/fail gate; it's a risk memo per finalist. "Low risk, mild amusement in Brazilian Portuguese, rhymes with nothing" is a perfectly shippable result. Perfection is not available; informed choice is.

Trademark classes are where names actually die

Folk wisdom says trademark disputes come from bullies. Reality: most collisions are same-vibes-same-class accidents between well-meaning founders who never checked. The discipline:

• Know the Nice classes that matter for you. Software lives in class 9 (downloadable) and 42 (SaaS). Retail brands live in 35. Food, hospitality, clothing all have their own. A name can be beautifully free in class 30 and radioactive in class 9 — which matters the day your coffee brand ships an app.
• Screen registers, not just the web. Search IP Australia's ATMOSS, the USPTO, EUIPO, and WIPO's Global Brand Database for exact and phonetically similar marks in your classes. Confusability runs on sound as much as spelling — this is the part founders consistently miss.
• Check company registers and app stores too. Unregistered trading names and store listings create practical (if not strictly legal) collisions. Your name next to a defunct but once-sued crypto project is a smell you can't rebrand away cheaply.
• Then hire actual counsel. The screen narrows twelve names to a safe three; the formal opinion protects the one you choose. When we took Copperline Mutual through naming, financial-services classes were the tightest part of the project — two beloved shortlist names died to existing marks in class 36, and it's why "hazard round comes before love" is engraved on our process.

Markets matter as much as classes. File in the markets you'll actually trade in within three years, not everywhere defensively — the Madrid Protocol makes multi-market filing sane, but every registration is an annuity, and registries increasingly expect genuine-use evidence.

The domain conversation (without the grief binge)

Let's settle the .com question with the same rule we set in the field guide: don't let the domain re-name the company. The international layer adds three refinements:

1. ccTLDs are a trust decision, not a technical one. In Australia and New Zealand, .com.au and .co.nz carry local legitimacy the .com can't buy; in Germany, .de is nearly mandatory for consumer trust. Budget for the home ccTLD and the .com or a clean gTLD as the canonical.
2. The parked-domain dance. A premium .com owned by an investor might cost more than your entire rebrand. The mature response is a modifier strategy (with, try, studio, hq) — but choose a modifier that's used consistently everywhere (email, socials, packaging), because inconsistency leaks type-in traffic constantly.
3. Email beats homepage. The domain you'll say aloud most is the one after the @. Optimise for spelling aloud on the phone: hyphens and numerals are permanent taxes on every support call your team will ever make.

Also secure the two or three obvious transliteration and typo variants in your primary market. Not the whole defensive stockpile — just the ones a confused five-year-old customer would type. Domain hoarding beyond that is an anxiety subscription.

Homophones, homographs and the spelling-bee problem

The tests that specifically exist because names travel:

• The accent test. Say the name in the three accents of your core markets. If it survives Australian vowels, a British clip and Singaporean pace, it will survive most of world English. Record yourself; the word you wrote and the word speakers produce are different animals.
• The heard-not-read test. Names spread by podcast ads, referrals and voice notes. Have a stranger hear the name once and type it into a browser. Their misspellings are your actual brand in the wild. "Wattle" and "waddle" were adjacent disasters waiting to happen when we named the restaurant group behind Wattle & Daub — we bought both domains on day one and never thought about it again.
• The script question. Roman-alphabet names travelling into Chinese, Japanese, Korean, Arabic or Cyrillic markets acquire transliterations whether you manage them or not. Better to choose the official rendering early — ideally one whose characters read well and mean nothing weird — than to let the market nickname you.
• The emoji and ASCII-art test. Frivolous until it isn't: will the name compress into a neat initialism? Does that initialism spell something? AUDI has four decent letters. Your four letters might spell a slur in a language you don't speak. One search per initialism; thirty seconds; do it.

The pre-flight checklist

Before any name is presented as a finalist, it carries a completed passport:

1. Desk screen across top-10 relevant languages — memo attached.
2. Native-speaker panel per tier-1 market — quotes on record.
3. Trademark screen in the classes that matter, registers link-logged.
4. Domain and handle availability mapped, acquisition costs estimated.
5. Accent test, heard-not-read test, script transliteration note.
6. A one-line "known risks" summary — stated by us, before the client finds them.

Total elapsed time: about two weeks alongside the creative work. Total cost: a rounding error against the rebrand a missed check causes. International naming isn't about finding a word with zero shadows; every real word has a shadow somewhere. It's about knowing where yours fall and deciding, eyes open, that your customers will never stand in them. That's the whole job — and it's learnable, procedural, and mostly just caring more than your competitors bother to.

Key takeaways

• Run linguistic screens in tiers: desk screen for everything, native-speaker panels after shortlisting, dialect and rhyme checks for finalists. Export risk memos, not verdicts.
• Trademark disaster happens in the classes, not the headlines. Screen registers phonetically, in the Nice classes that matter, before counsel formalises the survivor.
• Own ccTLDs where you trade; treat the exact-match .com as optional. Your most-spoken domain is the email one.
• Test homophones, accents, typing-after-hearing, transliteration into other scripts, and unlucky initialisms.
• Every finalist needs a passport: screens logged, risks stated, decisions documented. Two weeks, run in parallel with the creative.

FAQ

At what stage should we run the international checks? Tier 1 after generation (the 300-name pile, cheaply), tier 2 on the shortlist of 10–12, tier 3 and counsel on the final three. Checking everything exhaustively is burning money; checking nothing until you're in love is burning the house.

We only sell in one country. Can we skip this? Screens cost little; regret costs plenty. The internet leaks your name to every market on earth the day the site goes live, regardless of where you invoice. At minimum run tier 1, and spec the trademark classes with your future adjacent products in mind — app, merch, content, training. Class 9 surprises have ruined more launches than any translation error.

Are the famous mistranslation stories real? Mostly not, but the real history is interesting enough: the disasters that actually happened were usually pronunciation tone-deafness, register slips, or class collisions, not literal mistranslations. The pattern worth copying from the legends isn't "foreign languages are scary"; it's "someone with a clipboard should have checked".

What if our dream name fails one market check? Then the question stops being creative and turns strategic: does that market matter enough in the next five years to rename for? Sometimes the answer genuinely is no, and you ship with the risk documented and the local brand strategy adjusted. What you must not do is look away. An informed compromise is strategy; an ignorant one is an accident with a launch date.

Do we need different names per market for accessibility? Almost never. Consistent global naming with managed local transliterations is simpler and compounds brand equity across borders. The exception: regulated categories (finance, health) where local regimes sometimes require locally-filed trading names — a legal formality, not a rebrand.