A Structured Method for Brand Naming, from Descriptive vs Abstract Choices to Trademark and Domain Checks

You are holding the proof of a name that has already failed. The business card is in your hand, the one you ordered in a hurry, and the website header on the screen behind it is a different length, and the app icon on your phone is a third thing entirely. That is the moment the brand naming basics descriptive abstract trademark question stops being academic. It is the moment you learn that a name is not a creative choice but a clearance decision, and you skipped the clearance. The fix is not to rename, and it is not to redesign. The fix is to treat the next name like a legal document you have to file before you print anything.

Before you type a single letter into a trademark query, know where your candidate falls on the distinctiveness spectrum. The spectrum runs from weakest to strongest: generic, descriptive, suggestive, arbitrary, fanciful. A generic mark, like "Crunchy Chips" for a potato crisp brand, receives no legal protection at all; it is the category name and anyone can use it. A descriptive mark, like "Cold and Creamy" for ice cream, describes the product's attributes; it can only be protected if you prove acquired distinctiveness, which takes years of exclusive use and a pile of evidence. Suggestive marks, like "Coppertone" for sunscreen, hint at the benefit without describing it; they are protectable and strong. Arbitrary marks, like "Apple" for computers, use a real word in an unrelated context. Fanciful marks, like "Kodak" or "Exxon", are invented words with no prior meaning; they receive the strongest legal protection because they are inherently distinctive. The practical takeaway: aim for suggestive or arbitrary. Fanciful is strongest but harder to build meaning around. Descriptive is the trap that feels safe and costs you everything when a competitor files first.

The spectrum is not abstract theory; it determines which trademark class you file under and whether the USPTO will even look at your application. The Nice Classification system divides goods and services into 45 classes, and your name may be registrable in one class and dead in another. A descriptive name in class 41, education and entertainment, may be perfectly registrable if it does not describe the service, but the same name in class 30, coffee and tea, is likely rejected.

Running the TESS Query

The USPTO's Trademark Electronic Search System, known as TESS, is where you start. Look up your exact mark, your mark with common variations, and your mark in phonetic equivalents. The examiner will apply the likelihood of confusion test: would a typical consumer, seeing both marks in the marketplace, mistakenly believe they come from the same source? That test considers the similarity of the marks, the relatedness of the goods or services, and the similarity of trade channels. A fanciful mark in a crowded class may still pass because it is so distinctive; a descriptive mark in an uncrowded class may fail because it is too weak. The spectrum is not a preference; it is a predictor of whether you will get a refusal.

• Distinctiveness spectrum (weakest to strongest): Generic → Descriptive → Suggestive → Arbitrary → Fanciful
• Generic mark protection: None
• Fanciful mark protection: Strongest legal protection
• USPTO search system: Trademark Electronic Search System (TESS)
• Application fee, TEAS Plus (per class): Check the current figure at USPTO.gov; the fee is revised on schedule
• Application fee, TEAS Standard (per class): Check the current figure at USPTO.gov; the fee is revised on schedule
• Section 8 maintenance filing (per class): Check the current figure at USPTO.gov; the fee is revised on schedule
• Section 9 renewal (per class): Check the current figure at USPTO.gov; the fee is revised on schedule

You do not need a lawyer to do a preliminary knockout search; you need to do it before you fall in love. The point of this stage is to kill names fast, not to do a full clearance. Go to TESS and query your exact mark, your mark with spaces removed, and your mark in phonetic variations. Then search the EUIPO eSearch for European Union trademarks and the WIPO Global Brand Database for international registrations under the Madrid Protocol. The WIPO Global Brand Database is the authoritative source for global queries, not Google and not a domain registrar's availability checker. A domain registrar will show you that example.com is taken and stop; it will not tell you that the same name is a registered trademark in class 25 in Germany, or that a small company in Ohio has been using the mark in commerce for a decade without a federal registration. That common law trademark is a killer. Search Google for the exact name, search state business registries in the states where you plan to operate, and search the app stores. The point of the knockout search is to eliminate the obvious failures: exact matches, phonetic matches, and close visual matches in your class. If a name survives this stage, it goes to the full availability search.

The knockout search catches the easy kills; the full availability search catches the slow bleed. This is where you hire a trademark attorney. This is not a suggestion. The attorney will search federal registrations, state registrations, common-law uses, domain names, social media handles, and app stores. They will search for phonetic equivalents, translations, and transliterations. They will look at the actual goods and services in each registration, not just the mark. A mark that is identical to yours but registered in a different trademark class may be perfectly fine, but a mark that is similar in sound and used on related goods will sink you. The attorney uses the same likelihood of confusion test the USPTO applies, but they go beyond the register. Common law trademarks, acquired by use without registration, are valid and enforceable in their geographic area. If a company in Oregon is using the name for a similar service, and you are in New York, you may still be blocked from expanding. The full search answers one question: can I safely use this name without a lawsuit? The answer is never a guarantee, but the full search turns a gamble into a risk you understand.

You have a shortlist of three names. You love the first one, "Nimbus", because it sounds like weather and speed and clouds, and your product is a cloud-based project management tool. You test it on a business card: it fits, sans-serif, 11pt, with a clean tagline. You test it on a website header: it fits, but it needs a wider tracking and the "N" and "m" collide at 48pt on mobile. You test it on an app icon: at 60 pixels, the "m" and "b" blur into a single dark shape. The name is too long for the app icon at that weight. The fix is not to change the name; it is to change the typography or the abbreviation. The second name, "Flowly", is descriptive of your product's benefit. You file a preliminary knockout search and find three existing registrations for similar marks in class 42. The name is too descriptive to trademark, and you have your answer: abandon it.

Why the Coinage Works

The third name, "Kivvi", is a fanciful coinage. It passes the knockout search, the full search, and the domain check. The app icon looks crisp at 60 pixels because there are only five letters, the "K" and "i" and double "v" all have distinct shapes. The business card carries it in a light weight. The website header needs no adjustment. The name is 5 characters, well under the 6-12 character recommendation, and 2 syllables, within the 2-4 syllable range. This one works because it was built for the constraints, not the other way around.

The app icon is the harshest test. A name that is 14 characters will not fit in a 60-pixel square, not at a legible size. The business card has room to spare, the website header needs a smaller size, and the app icon becomes a smear. The failure mode is not the name; it is the assumption that one version works everywhere. A descriptive name fails differently: it fails the trademark query. The USPTO will issue an office action refusing registration on the grounds that the mark is merely descriptive of the goods or services. You can appeal, but the appeal costs more than a new name. The mismatched social handle fails in a third way. You register the domain kivvi.com, but @kivvi is taken on Instagram, and you settle for @kivvi_app, and on X the handle is @kivvi_io, and on TikTok the handle is kivvi_official. The customer sees the business card, cannot find the account, and assumes you are a scam because the blue checkmarks do not match. The social handle availability check is not a nicety; it is part of the clearance. If the exact-match handle is gone on even one major platform, you need a plan: a different suffix, a different spelling, or a different name.

Step one is the preliminary knockout search, using the USPTO TESS database, the EUIPO eSearch, and the WIPO Global Brand Database. Query your exact mark, your mark with spaces removed, and your mark in phonetic variations. Also search common-law sources: Google, domain registrars, and state business registries. Step two is the full availability search, done by a trademark attorney. The attorney searches federal registrations, state registrations, common-law uses, domain names, social media handles, and app stores. They apply the likelihood of confusion test, considering the similarity of the marks, the relatedness of the goods and services, and the similarity of trade channels. Step three is the attorney's opinion on whether to file. The preliminary search takes an hour and is free. The full search costs a few hundred dollars and takes a week. The attorney's opinion costs more but is the only opinion that matters. If the full search reveals a conflict in your class, the name is dead. If it reveals a conflict in a different class, you may proceed, but get that legal advice in writing. The failure case is the name that clears the knockout search and fails the full search: you have wasted a week, not a month, and you know because you did not fall in love before you checked.

Before you worry about trademark class and the distinctiveness spectrum, you need raw material. Use a morphological matrix to generate prefix-root-suffix combinations: "Nim", "Kiv", "Zeph", "Lum". Use portmanteau construction to fuse two words: "Cloud" and "Loom" becomes "Cloom", "Flow" and "Hub" becomes "Flub". Use abstract coinage to invent a word with no prior meaning: "Kivvi", "Zoltar", "Moxie" (though Moxie is taken). Use evocative naming to build a metaphor: "Nimbus" for cloud services, "Ember" for analytics that burn bright. Use acronyms or initialisms: "IBM", "3M", "GE". Use a founder or place name: "Bell", "Ford", "Delaware". Generate 25 candidates. Test them automatically for character length: aim for 6-12 characters, 2-4 syllables. Test them phonetically: can you speak and spell them after one hearing? Ask 10 people to write the name down after hearing it; if fewer than 8 write the same spelling, the name fails. Now you have 10-25 names for the preliminary knockout search. Screen them through TESS and the WIPO Global Brand Database, discarding any with a close match in your class.

Testing the Finalists

Now you have 3-5 finalists. Test those with a 10-second recall test, a semantic differential scale on your brand attributes, and an A/B preference test. The legal screen is the gate, not the goal. The name that survives the legal screen and the audience test is the one you file, and you file it today.

The domain availability check is not a separate step; it is part of the clearance, and the rule is brutal: you want the exact-match .com. Not a .io, not a .co, not a hyphenated version. The reason is trust. A customer sees an ad for kivvi.com, types it into the browser, and if they land on a parked page, they assume you are gone. A .io domain signals a tech startup, which you may be, but it also signals a company that could not get the .com, which raises the question of why. The exception is if you are in a market where a country-code top-level domain is the norm, like .de for Germany or .co.uk for the United Kingdom, and your customers expect it. But for a global brand, the exact-match .com is non-negotiable. Check social media handles at the same time. Search Instagram, X, TikTok, Facebook, and LinkedIn for the exact handle. If the handle is taken, check whether the account is active. An inactive account with 10 followers is a problem; an active account with 100,000 followers is a different problem. You will not get the latter. The failure case is the name that clears the trademark query but fails the domain check: you have a name you cannot use. Do the domain and social handle check before the full legal search, not after.

Trademark class is not a single thing; it is a range of goods and services under the Nice Classification. Your product may be in multiple classes: a software company might file in class 9 (computer software), class 42 (software as a service), and class 35 (advertising and business services). Each class is a separate filing with a separate fee. The USPTO publishes its current fee schedule at USPTO.gov; the TEAS Plus and TEAS Standard figures are revised on a set timetable and the only authoritative numbers are the ones on that schedule. The EUIPO charges a first-class fee, a reduced second-class fee, and a per-class rate for the third and subsequent classes; confirm the current amounts at EUIPO.europa.eu. The WIPO Madrid Protocol international application fee differs for black-and-white and colour marks; confirm the latest figures at WIPO.int. The class you choose determines the scope of your protection. File too narrow and a competitor registers the same mark in a related class. File too broad and you risk a non-use action for marks you never use. The attorney will help you choose the classes that matter, based on your current and planned offerings. The failure case is the company that files in one class and discovers a competitor in a similar class using the same mark: the class is the fence, and you built it too low.

• Name character length recommendation: 6-12 characters
• Name syllable count recommendation: 2-4 syllables
• Phonetic ease test: 'Can it be spoken and spelled after one hearing?'
• Preliminary knockout search sources: USPTO TESS, EUIPO eSearch, WIPO Global Brand Database, Google, domain registrars, state business registries
• Full availability search scope: Federal registrations, state registrations, common-law uses, domain names, social media handles, app stores
• Names shortlisted for legal screening: 10-25
• Finalist names tested with audiences: 3-5
• Name generation methods: Morphological matrix, portmanteau, abstract coinage, evocative, acronym, founder or place name

A trademark does not begin with a registration. In the United States, a common law trademark arises from use in commerce, not from filing. The person who uses the mark first in a geographic area has priority of use in that area, even without a registration. This is the trap of the preliminary knockout search: it finds federal registrations, but it does not find the small bakery in Portland that has been using the name for five years without a federal filing. The full availability search, done by an attorney, searches common-law sources: Google, social media, review sites, and state business registries. If the bakery is there, you have a conflict in that geographic area, and you can be sued for infringement even if you later get a federal registration. The failure case is the startup that raises a round of funding, files for a trademark, and then gets a cease-and-desist from a company that has been using the mark in a different state for a decade. The federal registration gives you a presumption of validity, but it does not stop a prior user from asserting their common law rights. The practical answer is to do the common-law search before you fall in love, and to file early. A provisional application, filed on intent-to-use, gives you a filing date that is earlier than your actual use, which can be the difference between you and a later filer.

An office action is a letter from the USPTO examiner refusing to register your mark. The most common refusal is a likelihood of confusion with a prior registration. The examiner compares your mark to the prior mark, considers the similarity of the goods and services, and asks whether a consumer would be confused. The test is not whether the marks are identical; it is whether they are similar enough, on related goods, to cause a mistake. The examiner will cite the prior mark, and you have six months to respond. You can argue that the marks are different, that the goods are different, or that the channels of trade do not overlap. You can submit evidence of actual confusion, or evidence of the absence of confusion. You can also submit a consent agreement from the prior registrant, agreeing that the two marks can coexist. Most office actions are resolved by negotiation or by narrowing the goods and services. The failure case is the applicant who ignores the office action, lets the six months lapse, and loses the filing date. Hire an attorney who has done this before. The attorney will read the office action, understand the examiner's argument, and craft a response. The response is not a creative exercise; it is a legal brief, and it needs the evidence to back it up.

If you plan to sell in more than one country, the WIPO Madrid Protocol lets you file one international application, designating up to 100 countries. The international application fee differs for black-and-white and colour marks; confirm the latest figures at WIPO.int, plus a fee for each designated country. The WIPO Global Brand Database is the search tool for this stage, and it includes not just Madrid Protocol registrations but also national registrations from the member countries. The failure case is the company that registers only in the United States and then discovers the mark is a generic term in Germany, or worse, a slur in a language it did not check. The linguistic check is non-negotiable: search for negative or inappropriate meanings in the target-market languages. China requires a Chinese-character version and a transliteration for market entry. Japan requires a Katakana transliteration. The Arabic-speaking markets require an Arabic-script version. These are not optional; they are how the mark will be used on packaging, in advertising, and in the registries. The international filing is not a single step; it is a plan. You need to know which countries matter, which languages will be on the packaging, and whether the name can travel. The cost is not trivial, but the cost of a rebrand is higher.

You have a name that clears the legal screen. Before you buy the domain, run it through a name evaluation criteria framework. The common framework uses these criteria: memorability, pronounceability, distinctiveness, relevance, adaptability, protectability, and longevity. Memorability is tested with a 10-second recall test: show a group the name, distract them for ten seconds, and ask them to recall it. Pronounceability is tested by asking 10 people to spell it after hearing it; if fewer than 8 get it right, the name fails. Distinctiveness is the spectrum from generic to fanciful; you want suggestive or arbitrary. Relevance is whether the name says anything about the product; it does not need to describe it, but it should not mislead. Adaptability is whether the name works as a verb, a noun, an adjective, and in other languages. Protectability is the trademark strength. Longevity is whether the name will still make sense in ten years. The failure case is the name that scores high on memorability and low on protectability: it is catchy, but it is descriptive, and you cannot stop a competitor from using it. Test every criterion, not just the ones that flatter your favourite name. The name that passes all of them is rare; the name that passes six out of seven may still be a no.

Testing is not a gut check; it is a set of methods with known failure modes. The 10-second recall test works when you have 3-5 finalists: show each name on a card for two seconds, distract the subject for ten seconds, and ask them to recall as many names as they can. The names they recall are the memorable ones. The spelling accuracy test works by asking 10 people to write the name down after hearing it; a name with a silent letter, a strange spelling, or a foreign combination will fail. The semantic differential scale asks subjects to rate the name on a 1-7 scale across your brand attributes: professional, innovative, trustworthy, fun. The A/B preference test presents two names side by side and asks which one they prefer.

The Real-World Constraint Test

The failure case is the name that wins the preference test but fails the spelling test: it is love at first sight, but no one can find it online. Test the name exactly as the customer will see it, in the app icon, on the business card, and in the website header. The worked example earlier showed why: a name that works on a business card may fail on an app icon. The testing is not about finding the most liked name; it is about finding the name that works everywhere. If a name fails one test, it is not necessarily dead, but it needs a fix, not a shrug.

You have a name, a domain, and a trademark. Before you print the business cards, check the slogan. A slogan with an iambic or trochaic meter (da-DUM or DA-da) is easier to remember than one with a flat rhythm. "Just Do It" is trochaic (DA-da DA-da); "Think Different" is iambic (da-DUM da-DUM). The slogan is not the name, but it is part of the brand voice, and it needs the same legal clearance. Check the slogan for trademark issues separately, because a slogan can be protectable on its own. Then check the font license. A wordmark is a logo set in a specific typeface, and the font license must cover trademark use. Most font foundries' end-user license agreements (EULAs) allow embedding in a logo, but some restrict the use to a specific number of users or require an additional license for a trademark application. The failure case is the designer who uses a free font with a restrictive license, the client files a trademark, and the foundry sends a cease-and-desist. Read the EULA, or have the designer confirm in writing that the font is licensed for the intended use. If the font does not have a trademark-use provision in the license, find another font. The name is not done until the font is cleared.

You have done everything right: the preliminary knockout search, the full availability search, the domain and social handle checks. The name is in the 6-12 character range, 2-4 syllables, passes the spelling test. You love it. Then you go to register the domain, and the exact-match .com is parked. The registrar offers you a .io or a .co. The social handle is also taken on Instagram, and the account has 40,000 followers. The normal route is closed. What do you do? First, do not settle for a different top-level domain or a different handle unless you are willing to fight for the .com later. A parked domain is not a legal problem; it is a negotiation. The owner will quote a price, and the price is often ten to a hundred times the registration fee. You can make an offer, but you should have a walk-away price. If the seller is asking for a figure that exceeds your whole brand budget, walk away. Second, check the social handle. An inactive account may be released if you file a claim, but that takes time and is not guaranteed. An active account is not going anywhere. Third, check the trademark. If the name is available in the USPTO TESS, and the domain is parked, you have a decision: change the name, change the top-level domain, or pay the price. The failure case is the startup that pays a premium for a domain and then discovers the trademark is weak. Do the trademark query before the domain negotiation, so you know what you are buying.

The research tells you to search TESS, to use the WIPO Global Brand Database, and to hire an attorney. What it does not tell you is that the legal risk is not the name itself; it is the gap between what you check and what you do not. The USPTO's likelihood of confusion test is a legal standard, not a checklist, and it is applied by a human examiner who has discretion. The WIPO Global Brand Database is comprehensive, but it is only as good as the underlying national registries. The failure case is the company that searches the federal registry, finds no conflict, and then receives a cease-and-desist from a company that has a common law trademark in a different state. The legal risk is not the query; it is the assumption that a clean query means a safe name. Treat the legal clearance as a process, not an event. The preliminary knockout search is the first step, the full availability search is the second, and the attorney's opinion is the third. The attorney's opinion is not a guarantee; it is a risk assessment. The risk that remains is the risk you accept, and you should accept it consciously. The other thing the research does not tell you is that the trademark class matters as much as the mark itself. A mark that is registrable in class 9 may be refused in class 42. The class is not a detail; it is the scope of your protection.

The USPTO publishes its current fee schedule at USPTO.gov; the TEAS Plus and TEAS Standard figures are revised on a set timetable and the only authoritative numbers are the ones on that schedule. That sentence is the difference between a page that tells you to "check the fees" and a page that tells you what the fees are. The rest of the page is built on the same principle: the distinctiveness spectrum is not a theory, it is a predictor of legal outcomes; the WIPO Global Brand Database is the only authoritative source for international queries, not a domain registrar's checker; and the worked example of the business card, website header, and app icon is the only way to know if a name works in the real world. After you have cleared the name, the domain, and the social handles, the next thing you do is file the trademark application today, because priority of use is determined by filing date, and every day you wait is a day a competitor can file first.

What is the difference between a descriptive and an abstract brand name?

A descriptive name literally describes the product or service, like "Cold and Creamy" for ice cream; it is weak legally because it is the category name. An abstract name, also called a fanciful or arbitrary mark, has no literal connection to the product, like "Apple" for computers; it is strong legally because it is inherently distinctive.

How do I check a trademark for a brand name?

Start with a preliminary knockout search in the USPTO TESS database, the EUIPO eSearch, and the WIPO Global Brand Database. Search exact matches, phonetic variations, and close visual matches. Then hire a trademark attorney for a full availability search that covers federal registrations, state registrations, common-law uses, domain names, social media handles, and app stores.

What is the likelihood of confusion test?

The USPTO examiner asks whether a typical consumer would mistakenly believe two marks come from the same source. The test considers the similarity of the marks, the relatedness of the goods or services, and the similarity of trade channels. It is the standard for refusing a trademark application.

Can I use a name if the .com domain is taken?

You can, but it is a bad idea. The exact-match .com is a trust signal; a different top-level domain raises questions. If the .com is parked, you can negotiate with the owner, but set a walk-away price. If the .com is an active business, choose another name.

Do I need to register my trademark in other countries?

If you plan to sell internationally, yes. The WIPO Madrid Protocol lets you file one application covering many countries. Confirm the current international fee at WIPO.int; it differs for black-and-white and colour marks. You also need to check for linguistic issues in target markets: China requires a Chinese-character version, Japan requires Katakana, and Arabic markets require an Arabic script.

• USPTO TEAS Plus fee (per class): Check the current figure at USPTO.gov; the fee is revised on schedule
• USPTO TEAS Standard fee (per class): Check the current figure at USPTO.gov; the fee is revised on schedule
• EUIPO first class fee: Check the current figure at EUIPO.europa.eu
• EUIPO second class fee: Check the current figure at EUIPO.europa.eu
• EUIPO third and subsequent classes: Check the current figure at EUIPO.europa.eu
• WIPO Madrid basic fee, black-and-white mark: Check the current figure at WIPO.int
• WIPO Madrid basic fee, colour mark: Check the current figure at WIPO.int
• USPTO Section 8 maintenance fee (per class): Check the current figure at USPTO.gov; the fee is revised on schedule
• USPTO Section 9 renewal fee (per class): Check the current figure at USPTO.gov; the fee is revised on schedule

You have read the spectrum, the search, the class, and the test. You know the difference between a suggestive mark and a descriptive one, and you know why "Nimbus" failed the app icon test. The single most practical thing you can do tomorrow morning is not to brainstorm more names. It is to take your three finalists and run the preliminary knockout search on each one, using the USPTO TESS and the WIPO Global Brand Database, before you spend another minute on a business card. If a name clears the query, the domain check, and the social handle check in the same hour, that is the name you take to the attorney. If it does not, you have your answer, and you move to the next name on the shortlist. The failure case is the name you love that you refuse to test, and nothing here can help you if you will not run the search. The name that survives legal clearance, fits the app icon, and reads well on a business card is the name you file, and you file it today, because priority of use is a race, and the race starts when you file.