How to Test a Brand Name Before You Commit to It

What a test can tell you (and what it can’t)

“Test your name” sounds obvious until you try to do it, because most of what people call testing is really opinion-collecting. The useful move is to split the question into what is measurable and what is not, and only spend testing effort on the first pile:

• Testable: functionWhether the name works as a word: sayable, spellable, memorable, ownable, and clean across the languages of your markets. These are properties of the string itself, so they are checkable today, before a single customer exists.
• Not testable: tasteWhether the name feels like you. Two candidates can pass every check below and still feel completely different to wear. Testing narrows the field to names that work; choosing among workable names stays a judgment call, and that is fine.
• Not covered: legalTrademark conflicts and social-handle availability. A name test is not legal clearance, and this guide will not pretend otherwise. Screen cheaply first, then spend counsel time on the one or two finalists that earned it.

Everything below stays inside the first row: observable behavior and checkable facts. The moment a test asks someone what they feel about a name, it has left science and entered small talk.

Why “I asked ten friends” is not a test

The default validation method, polling friends and colleagues, fails three ways at once. Friends are polite: they answer the question “do you like it?” with kindness, not prediction, and kindness is worthless as data. Ten opinions are statistical noise: nobody would ship a product on a sample of ten likes, yet founders bet their brand on exactly that. And novelty distorts everything: a brand-new name sounds slightly wrong to everyone, including you, for about a week, while a familiar name sounds right purely because it is familiar.

The fix is to stop measuring opinion and start measuring behavior. Can they spell it after hearing it once? Do they remember it tomorrow? Did they notice it in a lineup? Behavior predicts what the market will do with your name; opinion predicts how nice your friends are. Every test below is behavioral for exactly this reason.

The seven tests

Run them on any serious candidate. Each one targets a different failure mode, so a name must clear all seven, not average them; the order they appear here is the order to run them (the next section explains why).

1. The say-it testIntroduce your company out loud, in a full sentence: “Hi, I’m Ana. I run [Name].” Ten times, different rooms, different moods. A stumble you can feel in your own mouth becomes a stumble in every podcast ad, sales call, and investor intro. Names are spoken assets first; if the mouth resists, the market will too.
2. The spell-it-from-sound testSay the name once to someone who has never seen it written (over the phone is ideal) and ask them to type it. Count the corrections. Every correction is a future customer typing the wrong address and finding someone else. Self-spelling names, heard once and typed right, are the gold standard.
3. The day-after recall testTell five people the name today, in one normal sentence, with no coaching. Tomorrow, ask them what it was. Unaided recall after a day is the most honest memorability signal a founder can collect for free. What people misremember it as is data too: their mutation is sometimes the stickier name.
4. The lineup testWrite the name in a flat list with six real competitors and show it to someone for five seconds. Ask which one they noticed. Memorability is relative: a name that reads fine alone can vanish on a shelf of category clichés, and in a category full of Flow-and-Spark sameness, the odd one out wins.
5. The domain checkCheck the extensions your buyers actually trust: .com, .io, .app, .co. Treat every result as worth verifying rather than fact (registries move by the minute), and distrust any lookup that never answers “unknown”. A strong name with nothing ownable is a strong name you rent from someone else.
6. The cross-language screenSay the name in the accent of each target market, then search what that sound collides with in the local informal register. You are hunting slang, not dictionary entries. An honest screen reports “no flags found in the languages checked”; a tool claiming a name is “safe in every language” is telling you about itself.
7. The context testPut the name where it will actually live: an email signature, an invoice header, “the [Name] app”, a conference badge. Some names thrive in a logo and die in a sentence. Register matters too: playful syllables read charming on a consumer app and alarming on an audit tool.

Run them in order: cheap kills first

The sequence is the method. The say-it and spell-it tests take minutes, cost nothing, and kill weak names fast, so they go first. Recall and lineup need a day and a few volunteers, so they run on the survivors. Domain and language checks are free too, but there is no point checking ownability on a name your own mouth already rejected. The context test comes last because it is the slowest to read and the most subjective.

Expect casualties: a shortlist of ten candidates commonly loses half by the third test, and that is the system working. A name that dies on test two cost you ten minutes. The same flaw discovered after launch costs a rebrand, a redirect strategy, and every ounce of recognition you had built. The funnel exists to make failure cheap.

Score it, don’t vibe it

Impressions leak between candidates; numbers hold still. After the tests, force each name into a score per category on a 0 to 100 scale, and write a one-line justification next to every number. The justification is the honesty mechanism: “Fair (55): two spelling forks when heard aloud” can be argued with, while a bare “55” cannot. If you can’t write the line, the number is a vibe wearing a digit.

Namoly’s verdict bands make a useful reading frame: 80–100 is Strong, 60–79 Good, 40–59 Fair, 0–39 Weak. Weight the categories before comparing candidates, not after (deciding weights after seeing scores is how the favorite always wins). The weighting Namoly uses is a reasonable default: domain availability 0.30, memorability 0.25, phonetics 0.25, cross-language safety 0.20. Ownability carries the most weight because it is the one property you cannot iterate your way out of later.

What you still can’t self-test

A guide that claims its checklist is sufficient is selling something. Three things stay stubbornly outside any DIY battery, and the honest plan is to name them and budget for them:

• Trademark clearanceNo DIY screen replaces a proper search by counsel. Run it on the finalist or two, never on the whole list; the cheap tests above exist precisely so the expensive one happens once.
• A native speaker’s cultural readAutomated screens and informal-register searches catch the loud collisions. The quiet ones (tone, connotation, a decade-old local meme) need a human who lives inside the language. Book that read for the final pick only.
• TimeEvery new name sounds slightly wrong for the first week, and every familiar name sounds right only because it is familiar. Park the winner for a few days, then re-run the say-it test before committing. Adrenaline is not data.

The automated version of this page

Everything above that is a property of the string itself can be computed, and computing it is faster and less forgiving than doing it by hand. Namoly runs the four core checks (domain availability, memorability and distinctiveness, phonetics and readability, cross-language safety) deterministically: the same name returns the same scores for the same reasons, every time, with no dice-roll and no “regenerate and hope”.

Two details keep the automated version honest. Every category ships with a plain-English justification, a tip, and a confidence signal, so a shaky result never dresses up as a confident verdict. And the report states what it did not check (trademarks, social handles) right on its face. The efficient workflow: let the tool shorten your list in seconds, then spend the human tests (recall, lineup, context) on the two or three names that survived.

Frequently asked questions

How do I know if my brand name is good?

A good brand name passes function tests: people can say it after reading it, spell it after hearing it, recall it a day later, a credible domain is ownable, and it carries no flags in the languages of your markets. Score those and most “is it good?” anxiety resolves. What remains is taste, and taste is yours to spend.

Can I test a brand name without buying the domain?

Yes. Checking a domain’s status costs nothing and commits you to nothing; buying is a separate decision for later. Score the name first, then buy once, for the winner. Registering ten domains to “hold” a shortlist is the expensive version of a spreadsheet.

How many people do I need to test a name with?

Five cold readers beat fifty polite friends. You are measuring behavior (spelling, recall, pronunciation), not collecting votes, and behavioral failures surface fast: if three of five people misspell the name, a sixtieth opinion will not save it. Prefer people who resemble your buyers and have no stake in your feelings.

Should I run a naming survey or focus group?

Carefully, if at all. Groups reward safe, familiar-sounding names and punish distinctive ones, which is the opposite of what a brand needs. If you must survey, measure behavior (unaided recall, spell-from-sound, a forced choice inside a competitor lineup) instead of asking “which do you like?”.

Does a brand name test cover trademarks?

No. Nothing on this page, and nothing in Namoly’s report, is trademark or social-handle screening; the report says so on its face. Use the tests to shortlist, then put the finalist in front of counsel before you commit money to it.

How long does testing a brand name take?

An afternoon by hand for the full battery, plus a day of waiting for the recall test; the automated four checks take seconds. Compare that with the alternative: an untested name that fails in public costs a rebrand, and rebrands are priced in months, not afternoons.