How Brands Get Named

Someone says it in almost every naming conversation I have sat in, usually early, usually as a warning: the Chevrolet Nova failed in Latin America because no va means "it doesn't go" in Spanish. It is the origin myth of linguistic screening, and it is false.

Snopes rates the claim false and gives two reasons. The first is phonetic: no va and nova have different pronunciations in Spanish, stressed differently, and a Spanish speaker no more hears one in the other than an English speaker hears "no table" in "notable". The second is commercial: the Nova sold well in both of its main Spanish-language markets, Mexico and Venezuela, and Venezuelan sales actually exceeded General Motors' expectations. GM looked at the translation, decided it did not matter, and kept the name. A related detail in the legend, that GM renamed the car the Caribe, is also wrong; the Caribe was made by Volkswagen.

I keep returning to this because the myth teaches the wrong lesson in a memorable form. It suggests naming risk is about embarrassing translations, which is the smallest and most easily managed risk in the whole process. The real work is legal, procedural and slow, and almost none of it is about whether a word sounds funny somewhere else.

What the law is actually sorting for

The framework every trademark lawyer uses comes from one opinion: Judge Henry Friendly writing for the Second Circuit in Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4 (2d Cir. 1976). The passage is short and it has organised half a century of practice:

arrayed in an ascending order which roughly reflects their eligibility to trademark status and the degree of protection accorded, these classes are (1) generic, (2) descriptive, (3) suggestive, and (4) arbitrary or fanciful.

What each level buys is different in kind, not just degree. On generic terms Friendly is absolute: "no matter how much money and effort the user of a generic term has poured into promoting the sale of its merchandise and what success it has achieved in securing public identification, it cannot deprive competing manufacturers of the product of the right to call an article by its name." A merely descriptive mark can be registered only once it has "become distinctive of the applicant's goods in commerce", which in practice means years of evidence. And "If a term is suggestive, it is entitled to registration without proof of secondary meaning." Suggestive is the lowest rung where a name works on the day you file it.

Friendly also warns that "the lines of demarcation... are not always bright", and that a term's category can shift depending on the product, on how it is used, and over time. That caveat does more work than the categories themselves, and it is the part naming decks always drop.

The US Patent and Trademark Office publishes its own version with examples, which is useful because these are the office's own choices rather than a blog's:

The tension this creates is the central fact of commercial naming. The name a founder wants is usually descriptive, because it explains the business. The name a lawyer wants is usually fanciful, because it is defensible. WIPO's own guide for small businesses states the trade-off without flinching: "The legal strength of a trademark is often inversely related to the appeal of the trademark to your marketing team."

A subject catalogue. Classification systems decide what can be found, which is also what a trademark class does. Photograph by Dr. Marcus Gossler, CC BY-SA 3.0, via Wikimedia Commons.

Classes, and why "is it taken?" is the wrong question

Trademarks do not exist in general. They exist for specified goods and services, and the specification uses a system almost nobody outside the profession has heard of.

The Nice Classification was established by the Nice Agreement of 1957 and is administered by WIPO. It divides everything into 45 classes: 34 for goods (classes 1 to 34) and 11 for services (classes 35 to 45). The current text is the Thirteenth Edition, version 2026, in force from 1 January 2026. Contracting states are obliged to include class numbers in their official registration documents.

The headings are broad, and reading a few makes the point better than any explanation. Class 25 in full is "Clothing, footwear, headwear." Class 35 is "Advertising; business management, organization and administration; office functions." Class 16 covers "Paper and cardboard; printed matter; bookbinding material; photographs; stationery and office requisites, except furniture" and more. Class 9 runs to a paragraph beginning "Scientific, research, navigation, surveying, photographic, cinematographic, audiovisual, optical, weighing, measuring, signalling, detecting, testing, inspecting, life-saving and teaching apparatus and instruments".

So a studio asking whether a name is "taken" is asking an unanswerable question. Taken for what? The same word can be registered by different owners in class 25 and class 42, and frequently is. The question that has an answer is whether a confusingly similar mark exists for related goods or services, and that requires deciding your classes first. Deciding classes is a commercial decision disguised as an administrative one: it fixes what the business is allowed to become without filing again.

One thing I could not resolve. WIPO's main classification page states that "A new edition is published every three years and, since 2013, a new version of each edition is published annually", while WIPO's own Nice FAQ says a new edition appears every five years, with the same annual versioning. Both are current WIPO pages. I have not found which is the operative statement, and I mention it because it is a good illustration of how much of this material is secondhand even when it comes from the primary source.

What a clearance search actually checks

A clearance search is not a search for the exact word. The USPTO's published grounds for refusal make clear what the examining attorney is looking for, and only one of them is about identity.

On likelihood of confusion, registration is refused where the applied-for mark and a registered or pending mark are such that "the marks are similar and the goods and or services related such that consumers would mistakenly believe they come from the same source". Similar, not identical. Related, not the same. On descriptiveness: "The examining attorney will refuse registration of a mark as merely descriptive if it immediately describes an ingredient, quality, characteristic, function, feature, purpose or use of the specified goods or services." The other published grounds are geographic descriptiveness and geographic deceptive misdescriptiveness, being primarily merely a surname, and ornamentation, the last of which catches a surprising number of apparel brands whose mark only ever appears as a big graphic across the chest.

A search also has to cover more than one register. WIPO's Global Brand Database aggregates international marks under the Madrid System, appellations of origin and geographical indications under Lisbon, state emblems under Article 6ter, International Nonproprietary Names from the World Health Organization, and trademarks from participating national and regional offices. WIPO still advises that although the database "covers a number of large trademark collections, it may be prudent to search also the registers of national/regional intellectual property offices". On top of that sit unregistered rights, which do not appear on any register at all and are found by looking at the market rather than the database.

The sequence that survives contact with reality looks roughly like this. WIPO's own checklist for small businesses, in its guide Making a Mark, supplies most of it verbatim:

WIPO's checklist for selecting a trademark

1. Check to ensure no one else has registered the trademark or a
   confusingly similar one.
2. Check that the proposed sign meets all the absolute legal
   requirements for registration.
3. Do your best to select a strong trademark. The legal strength
   of a trademark is often inversely related to the appeal of the
   trademark to your marketing team.
4. Avoid imitating existing trademarks. A slightly altered
   competitor's trademark or a misspelled well-known or famous
   trademark is unlikely to be registered.
5. Consider possible limitations on registering a trademark which
   includes geographic words or signs.
6. Ensure that the trademark does not have any undesired
   connotations in your own language or in any of the languages of
   potential export markets.
7. Check whether the corresponding domain name (i.e., Internet
   address) is available for registration.
8. Make sure the trademark is easy to read, write, spell and
   remember and is suitable for all types of advertising media

WIPO's checklist for selecting a trademark

1. Check to ensure no one else has registered the trademark or a
   confusingly similar one.
2. Check that the proposed sign meets all the absolute legal
   requirements for registration.
3. Do your best to select a strong trademark. The legal strength
   of a trademark is often inversely related to the appeal of the
   trademark to your marketing team.
4. Avoid imitating existing trademarks. A slightly altered
   competitor's trademark or a misspelled well-known or famous
   trademark is unlikely to be registered.
5. Consider possible limitations on registering a trademark which
   includes geographic words or signs.
6. Ensure that the trademark does not have any undesired
   connotations in your own language or in any of the languages of
   potential export markets.
7. Check whether the corresponding domain name (i.e., Internet
   address) is available for registration.
8. Make sure the trademark is easy to read, write, spell and
   remember and is suitable for all types of advertising media

WIPO's checklist for selecting a trademark

1. Check to ensure no one else has registered the trademark or a
   confusingly similar one.
2. Check that the proposed sign meets all the absolute legal
   requirements for registration.
3. Do your best to select a strong trademark. The legal strength
   of a trademark is often inversely related to the appeal of the
   trademark to your marketing team.
4. Avoid imitating existing trademarks. A slightly altered
   competitor's trademark or a misspelled well-known or famous
   trademark is unlikely to be registered.
5. Consider possible limitations on registering a trademark which
   includes geographic words or signs.
6. Ensure that the trademark does not have any undesired
   connotations in your own language or in any of the languages of
   potential export markets.
7. Check whether the corresponding domain name (i.e., Internet
   address) is available for registration.
8. Make sure the trademark is easy to read, write, spell and
   remember and is suitable for all types of advertising media

Note where the language check sits: sixth of eight, after four legal steps. That ordering is correct and it is the opposite of how naming is usually presented.

The scale problem

The reason the first good name fails is arithmetic. WIPO's World Intellectual Property Indicators puts global trademark filing in 2024 at an estimated 11.7 million applications, covering an application class count of 15.23 million. Around 93.2 million trademarks were in force worldwide, up 6.1 percent on the previous year, with an estimated 8.3 million new registrations recorded in the year. Five offices account for 61.5 percent of filing by class count: China at roughly 7 million classes, the United States at 795,337, the Russian Federation at 567,227, India at 555,613 and Brazil at 468,667.

Ninety-three million live marks is the number to hold onto. Any pronounceable two-syllable English word, and most invented ones that look like English, is already registered somewhere by someone for something. The work is not finding an unclaimed word. It is finding a word whose existing claims sit far enough away from your classes.

Domains and handles

The domain layer has its own arithmetic. Verisign's Domain Name Industry Brief reported 386.9 million domain name registrations across all top-level domains at the end of the fourth quarter of 2025, of which 161.0 million were.com and 12.5 million.net, with country-code domains accounting for 145.6 million.

The practical consequence for a small studio's client is that the exact-match.com is gone, and has been for any real word since roughly the middle of the last decade. What replaces it is a set of compromises that are worth ranking honestly: a modifier before or after the name in.com, a country or category top-level domain, a deliberately invented word, or buying the domain from its current holder. The last is a budget line, not a formality, and it should be raised before the name is presented rather than after it is loved.

Handles are worse, because they are per-platform, non-transferable, first-come, and outside any legal framework a trademark gives you until a platform chooses to act. A name that clears legally and has a workable domain can still be unusable if the handle is held by a dormant account from 2011. I now check handles at the same point as the domain, not later, and I treat a name that fails on handles across three platforms as a name to drop.

Why the good name is usually not the first one

Put the stages together and the funnel is steep. A longlist has to survive: descriptiveness, which kills most of the names clients like; the likelihood-of-confusion check against registered and pending marks in the chosen classes; unregistered use in the actual market; the surname, geographic and ornamentation grounds; connotation in the languages of intended markets; a domain; handles; and finally whether anyone can spell it over the phone. Each stage is independent, and each one removes names at a rate that has nothing to do with quality. The name that emerges is not the best name. It is the best name that cleared.

There is one more reason not to grieve the loss of the obvious name: winning too completely is also a risk. In Elliott v. Google, Inc. the Ninth Circuit considered whether GOOGLE had become generic through public use as a verb. The court set out the governing test: "If the relevant public primarily understands a mark as describing 'who' a particular good or service is, or where it comes from, then the mark is still valid. But if the relevant public primarily understands a mark as describing 'what' the particular good or service is, then the mark has become generic." It held that a genericness claim must be directed at a particular type of good or service, so framing the question around the act of searching rather than internet search engines was improper, and affirmed summary judgment for Google. The opinion cites aspirin, cellophane and thermos as marks that did not survive the same test.

Genericide is the failure mode of a name that worked perfectly. It is also the clearest evidence that a name is not a fixed object. It is a legal position that has to be maintained, in a classification system, in specific markets, against specific neighbours, for as long as the business wants to keep it.

Sources

• Snopes, "Chevrolet Nova / No Va". Fact check rating the Nova translation legend false; the phonetic distinction between no va and nova; sales in Mexico and Venezuela; the Volkswagen Caribe confusion. https://www.snopes.com/fact-check/chevrolet-nova-name-spanish/

• Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4 (2d Cir. 1976), Friendly J. The four categories in ascending order; the passage on generic terms; descriptive terms requiring distinctiveness in commerce; suggestive marks registrable without secondary meaning; "the lines of demarcation... are not always bright". https://www.courtlistener.com/opinion/336319/abercrombie-fitch-company-v-hunting-world-incorporated/

• United States Patent and Trademark Office, "Strong trademarks". The fanciful, arbitrary, suggestive, descriptive and generic categories with USPTO's own examples, and the statement that descriptive marks are "only registrable in certain circumstances". https://www.uspto.gov/trademarks/basics/strong-trademarks

• United States Patent and Trademark Office, "Possible Grounds for Refusal of a Mark". Likelihood of confusion, merely descriptive and deceptively misdescriptive, geographic grounds, primarily merely a surname, ornamentation. https://www.uspto.gov/trademarks/additional-guidance-and-resources/possible-grounds-refusal-mark

• WIPO, Nice Classification. Established by the Nice Agreement (1957); publication cadence; obligation on contracting offices to publish class numbers. https://www.wipo.int/en/web/classification-nice

• WIPO, Nice Classification FAQ. "There are 34 classes of goods and 11 classes of services"; the conflicting statement on edition frequency. https://www.wipo.int/en/web/classification-nice/faq

• United States Patent and Trademark Office, "Nice Agreement current edition version: general remarks, class headings and explanatory notes". Nice Classification Thirteenth Edition version 2026 (NCL 13-2026), effective 1 January 2026; 45 classes with goods in 1 to 34 and services in 35 to 45; class headings for 9, 16, 25 and 35. https://www.uspto.gov/trademarks/trademark-updates-and-announcements/nice-agreement-current-edition-version-general-remarks

• WIPO, Global Brand Database. Coverage of Madrid System marks, Lisbon appellations and geographical indications, Article 6ter emblems, WHO International Nonproprietary Names and participating national and regional offices, with the advice also to search national registers. https://www.wipo.int/en/web/global-brand-database

• WIPO, Making a Mark: An Introduction to Trademarks for Small and Medium-sized Enterprises, Intellectual Property for Business Series No. 1. The eight-point checklist for selecting a trademark, quoted in full. https://www.wipo.int/edocs/pubdocs/en/wipo-pub-900-1-en-making-a-mark-an-introduction-to-trademarks-for-small-and-medium-sized-enterprises.pdf

• WIPO, World Intellectual Property Indicators 2025: Highlights, trademarks chapter. 11.7 million applications in 2024; application class count of 15.23 million; 93.2 million marks in force; 8.3 million registrations; the top five offices and their share. https://www.wipo.int/web-publications/world-intellectual-property-indicators-2025-highlights/en/trademarks-highlights.html

• Verisign, Domain Name Industry Brief, fourth quarter 2025. 386.9 million domain registrations across all top-level domains; 161.0 million.com and 12.5 million.net; 145.6 million country-code registrations. https://dnib.com/articles/the-domain-name-industry-brief-q4-2025

• Elliott v. Google, Inc., No. 15-15809 (9th Cir. 2017). The who-are-you/what-are-you test; the requirement that genericide be assessed against a particular type of good or service; the primary significance test under 15 U.S.C. 1064(3); affirmance of summary judgment; aspirin, cellophane and thermos as prior examples. https://cases.justia.com/federal/appellate-courts/ca9/15-15809/9129034504/0.pdf

Related reading

• How Studios Lose Their Own Work

• Pricing Creative Work

• Munich 1972 Was Public Infrastructure

More from the Moc Mallen Studio journal