What Are the 7 Types of Brand Names?

Naming practitioners commonly list seven types: descriptive, suggestive, invented, arbitrary real words, founder names, acronyms and compound or blended names. The type predicts the test risk. Descriptive names are easy to understand but hard to register, while invented names are easier to own but need recall and pronunciation testing because they start with no meaning.

Is a Free Brand Name Test Worth Running?

A free poll is worth running only as an internal gut check, never as the decision. Most free polls go to your own contacts, who know the context and share your language, so they miss the associations and misreadings a cold audience brings. Use one to cut a long list, then test the shortlist on a screened sample.

Should Product Names Be Tested Differently From Company Names?

Yes. A product name sits under a parent brand, so test it with the parent brand visible and measure whether it signals the right product and price tier. A company name has to carry the whole organization for years, which justifies more markets, more languages, a delayed recall task and a full legal screen.

What Is Adjectival Analysis in Name Testing?

Adjectival analysis shows separate groups the same product or ad under different names, asks an open question such as "What do you think?", then compares the adjectives each group uses. Because only the name changes between groups, the differences show which expectations each name creates. It tests meaning rather than preference.

How Many Names Should Go Into a Consumer Test?

Three to six names is the practical range after legal and linguistic screening. Fewer leaves no fallback if the favorite fails a late trademark check, and more multiplies the monadic cells, since each name needs its own sample of around 200 per market. A longer list belongs in a cheap screening round first.

Brand Name Testing Research Methods 2026

Last updated: 23 September 2026

Quick Answer: Brand name testing screens candidate names for trademark and linguistic risk, then tests the survivors one name per respondent for meaning, recall and pronounceability. Liking comes last, because on first exposure it leans on ease of processing. Budget about 200 respondents per name per market, in every language the name will be spoken.

In 2006 two Princeton psychologists found that newly listed NYSE companies with easy-to-pronounce names outperformed hard-to-pronounce ones after a day of trading, while across all 89 stocks the link was no longer statistically reliable at six months or a year. The study in PNAS put the day-one gap between the ten most and ten least fluent names at $112 on a $1,000 investment.

Ease of processing is real, but other information soon competes with it. A name test that leads with "which do you like best" is measuring the part of a name's effect that is least reliable over time. Whether the name can be owned, said aloud in every market and remembered with the right meaning attached does not decay, and that is what a brand or product name test should be built to read.

Which Research Methods Does Brand Name Testing Use?

Six methods cover the job, and a serious study uses at least four of them in sequence. Screening comes first, then association, then a monadic wave with recall and pronunciation tasks built in.

Legal and linguistic screening. A trademark search in every market and class, plus a native-speaker check of meaning, slang and sound in each language. No consumer sees a name that fails it.
Unprompted association. Show the name alone and ask what kind of company or product it suggests and what it would cost.
Monadic rating. Each respondent sees one name and rates fit with the positioning, distinctiveness and appeal. It follows the same one-per-respondent logic as monadic concept testing, so no name borrows context from another.
Delayed recall. Show the name, run an unrelated task for several minutes, then ask for it back unaided and ask the respondent to spell it.
Hear-and-spell. Play the name as audio and ask the respondent to write it down, then read it aloud from text and record how they say it. This catches names that fail in word of mouth.
Positioning sort. Give respondents three or four positioning statements and ask which name fits each. A name that sorts evenly across all of them carries no meaning yet.

The table sorts alphabetically by method.

Method	What it tells you	What it misses
Delayed recall	Whether the name survives a gap and can be spelled	Why a name was forgotten
Hear-and-spell	Whether spoken and written forms match	Meaning and fit
Legal and linguistic screen	Whether the name can be owned and said safely	Any consumer reaction
Monadic rating	Fit, distinctiveness and appeal without comparison bias	Recall over time
Positioning sort	Which meaning the name carries	Whether that meaning is liked
Unprompted association	The product and price people expect	Statistical confidence

Which Companies Run Brand Name Testing Research?

Four kinds of firm run name research. Naming agencies create, screen and often validate names, self-serve survey platforms run fast quantitative reads, research agencies run managed studies across markets, and trademark counsel give the legal opinion that no consumer study can replace.

Naming agencies. Catchword lists more than 1,500 naming and branding projects and sells trademark, linguistic and validation research alongside naming. Brand Institute states it has created more than 6,000 brand names and runs online naming research surveys benchmarked to historical norms, plus regulatory name-safety research. If you need names invented as well as tested, an agency is the right buyer.
Self-serve platforms. SurveyMonkey's name testing product takes up to 10 names, shows each respondent one name at a time, and returns results in as little as an hour from a panel it puts at 335 million people. Conjointly's brand name test scores each name on likeability, trust and how easy it is to remember and pronounce, and asks what comes to mind first. For a directional read on a product-line name within a day or two, either is faster and cheaper than a managed study.
Managed research. Full-service firms design the study, recruit, field and interpret it across markets. Alchemic fits here, offering end-to-end consumer research at scale: the name wave runs as adaptive interviews that ask why a name suggests what it does, on web, inside WhatsApp or by phone, with managed fieldwork or bring your own sample. It also offers a self-serve mode for teams that field their own reads.
Trademark counsel. Only counsel can give a clearance opinion, and only an examining office can grant registration.

What Does a Name Need to Clear Before Consumers See It?

A name has to pass a trademark screen and a linguistic screen before any consumer rates it. Testing first and screening second is the most expensive order available, because the team falls for a name it cannot register and the study has to be rerun on the runners-up.

The legal bar is wider than an exact match. The USPTO's guidance on likelihood of confusion calls it the most common reason for refusing registration, and says marks can conflict through similar sound, appearance, meaning or commercial impression, when the goods or services are related. It also notes that identical names can coexist when the goods are unrelated, which is how different owners register Dove for soap and for ice cream bars.

The register is also crowded. WIPO's World Intellectual Property Indicators 2025 estimates 11.7 million trademark applications filed worldwide in 2024, classified at many offices under 45 Nice classes of goods and services, so every candidate needs a search in each market and class it will trade in.

The linguistic screen asks native speakers in each market three questions:

Does the name mean anything, including slang, in this language?
Can a speaker say it without guessing, and does it sound like a word with an unwanted meaning?
Does the written form survive the local script and keyboard?

Why Does Liking Mislead in a Name Test?

Liking a new name leans heavily on how easily it is processed on first exposure, and processing ease fades once people learn what the name stands for. A liking score is worth collecting, but it belongs at the end of the questionnaire and at the bottom of the decision, below meaning, recall and pronounceability.

The Princeton result is the cleanest evidence. In the same study, lab participants with no other information predicted that fluently named stocks would rise and disfluently named ones would fall over a year. Across all 89 real stocks the effect held at a day and a week and was no longer reliable at six months, as the authors predicted, since other information comes to drive prices. The ten-versus-ten gap kept widening, so what fades is the reliability of the name signal, not its direction. A new name meets consumers in exactly that no-information state during a test, and in a very different one a year after launch.

Meaning does more durable work. In a 2017 study in Cognitive Research: Principles and Implications, Preziosi and Coane built 100 invented words designed to sound large or small, had participants define them as if they were products, and then tested memory for the pairings. Definitions that matched the sound of the invented word were recalled better than mismatched ones. A name that sounds like what it names is easier to remember, which is why the positioning sort and delayed recall matter more than a preference vote.

The same trap exists for visual identity, where familiarity rather than fluency does the damage; testing logos with consumers covers that case separately.

How Many People and Which Markets Does a Name Test Need?

Plan on about 200 respondents per name in each market, and a separate cell in every language the name will be spoken in. At 200 per cell, two names need to differ by roughly 10 percentage points on a yes or no measure before the gap is real at 95% confidence.

Why Every Language Needs Its Own Cell

Sound symbolism is widely shared across languages but not uniform. A study across 25 languages published in 2022, with 917 participants in nine language families and ten writing systems, found that the invented word "bouba" was reliably matched to a round shape in 22 of the 25 languages, while "kiki" was reliably matched to a spiky shape in only 11. Speakers of Romanian, Mandarin Chinese and Turkish gave matching answers less than half the time overall. If a sound-meaning link that basic splits by language, a coined brand name will too; an English-only test says nothing about the Tamil or Turkish reading.

Who a Browser-Only Name Test Leaves Out

Many names are heard before they are read: on the radio, at a shop counter, in a voice note between friends. A browser panel usually presents the name as text, and it reaches only people who complete surveys through links. Buyers who meet a name mostly by ear, including many with limited literacy, are the ones it misses.

WhatsApp interviews reach those respondents with no link and no app, and respondents can answer in voice notes, which turns pronunciation into data. AI phone interviews play the name down the line and record how it comes back. Alchemic publishes 57+ languages including Hindi, Tamil and Telugu, and fields across 14 markets including the USA and the UK, with India delivery from metros and Tier 1 through Tier 2 and Tier 3. The mechanics of running one study in several languages are in multilingual AI-moderated interviews.

How Do You Read a Split Result Between Two Names?

A split result, where two names tie or each wins different measures, is decided by the measures the name has to carry, not by averaging every score. Rank the measures before fielding, then apply them in order, and check every gap against the margin of error first.

Discard gaps inside the margin. At 200 per cell, a 6-point lead on a top-box appeal score is noise. Treat the names as tied on that measure.
Look for a disqualifier in the open ends. One unwanted association, repeated by a meaningful share of respondents in one market, outweighs a small lead everywhere else.
Apply the ranked measures. Meaning and fit with the positioning first, then recall and spelling, then pronounceability if the name will travel by word of mouth, then appeal.
Never pool markets. A name that wins in the US and fails in Brazil has failed, and a pooled score hides it.
Break a true tie on ownership. If the consumer data cannot separate two names, take the one counsel rates as more distinctive and easier to defend.

The winning name becomes the first asset a brand tracker measures. Setting the name test's measures to match the ones brand equity tracking will use later gives the launch a baseline instead of a guess.

Where Does Name Research Stop Being Useful?

Name research reduces the risk of a name that confuses, offends or is forgotten. It does not pick a great name, and four things stay out of reach however large the sample.

It cannot clear a trademark. The examining attorney searches the register whatever the research said.
It measures first exposure. Names are judged at their least familiar, so a distinctive but unusual name can score below a safe one that will be forgettable in a year.
It cannot create names. A study ranks the shortlist it is given. If every name on it is weak, testing picks the least weak.
Some modes carry no audio signal. Text interviews and text surveys cannot test pronunciation, so a study that needs it has to include a voice or phone mode.

Not every name justifies a managed study. A line-extension name inside an established brand rarely needs more than a self-serve monadic read, and the managed route earns its cost on company names, master brands and names that will launch in several languages at once. The wider rebrand sequence is in market research for branding.