How to Choose Between Your Final Name Candidates

Choosing between final name candidates is a different task from generating them, and scoring is bad at it. Across nine full Brand Cleared runs, our four-panel audit — a brand strategist, an industry expert, a growth marketer and a design director, each scoring independently — picked a name that did not out-score its own runner-up on the composite in three of nine runs (twice it scored lower, once it tied), and in four of nine the two finished within two points. A score removes names that are clearly worse. It does not separate two names that are both good.

So the useful question is not “which name scores highest?” It is “what should decide, once the scores stop deciding?” What follows is the order we’d run it in: hard constraints first, then the mouth, then the one dimension your business competes on.

Why doesn’t the highest-scoring name always win?

Because a composite score is an average, and averages hide the thing you are actually choosing between. A name that is 70/100 on every dimension and a name that is 95 on distinctiveness and 45 on pronounceability can land within a point of each other, and they are not remotely the same decision. The average tells you they are equivalent. They are not.

This is visible in our own runs. The winner-to-runner-up gap across nine full pipelines was −4, −3, 0, 1, 1, 2, 8, 10 and 12 points. Three of those are negative or zero — the panel deliberately declined to follow the ranking — and in four of the nine the two names sat within two points either way, which on a 100-point composite built from twelve weighted dimensions is inside the noise. Only three runs produced a gap of eight points or more, wide enough that the score alone was a defensible answer.

The practical reading: use scoring to get from a hundred candidates to five. Then put the scores down.

What should disqualify a name before you compare anything?

Hard constraints come first, because no amount of preference survives them and every one of them is cheaper to find now than after a rebrand. In order:

• A live trademark conflict in your classes. A confusingly similar registered or pending mark in the Nice classes you will operate in is a legal problem, not a branding preference. A knockout search on the USPTO register is the floor, and whether a specific conflict creates real exposure is a judgment for a trademark attorney.
• An existing business actually trading under the name. U.S. rights arise from use, not registration, so a company with no filing at all can still hold common-law rights that block you. This is why clearing a name means searching the open web, app stores and social platforms — not just the register.
• Nobody can spell it after hearing it once.Test it, don’t assume it. This kills more otherwise-good names than any other constraint.
• It breaks in a market you plan to enter. Check the meaning and the pronunciation in every language your first three years will touch.

Note what is not on that list: the .com being registered. A domain someone bought to resell is a price, not a conflict — nobody trades under it, nobody holds rights in it, and your business name does not have to match your domain. Letting a parked domain kill a name is the most common way founders talk themselves out of their best candidate.

How do I break a tie between two good names?

Take them out of the spreadsheet and put them in the four places the business will actually live. Ties almost always break in the first two:

1. Say it on a phone call.“Hi, this is Maria from ______.” If you find yourself spelling it, or adding “like the word, but with a K,” that is a permanent tax on every introduction the company ever makes.
2. Type it as an email address. maria@______.com. Double letters at the seam, silent letters and homophones all become obvious here and are nearly invisible on a shortlist.
3. Write it at the top of an invoice. Some names carry authority and some do not, and an invoice is where you find out.
4. Say it in the room you want to be in. A pitch, a shelf, a hospital corridor, a job offer. A name that is delightful in one of those can be disqualifying in another.

Should the name be descriptive, suggestive, or coined?

This is the question that decides the most and gets asked the least. The three categories trade off protectability against explanation cost:

• Descriptive (Northeast Dental Supply) — instantly clear, cheapest to market, and the weakest trademark position there is. Purely descriptive marks are refused federal registration without proof of acquired distinctiveness, and your competitors can describe themselves the same way.
• Suggestive (Netflix, Salesforce) — hints at the benefit without stating it. Protectable, and it still does some explaining for you. This is where most good startup names land.
• Coined (Kodak, Verizon, Xerox) — invented, therefore the strongest trademark position and by far the easiest to clear. You pay for it in the marketing budget that has to install the meaning.

The choice follows from your distribution, not your taste. If you will be found by search, a name that carries some meaning earns its keep. If you will be introduced by a salesperson or a founder, coined costs less than it looks and clears far more easily.

What actually decides, once the field is level?

Pick the single dimension your business genuinely competes on and let it break the tie. Not all twelve — one.

• Competing on trust — health, legal, financial, infrastructure — weight gravity and pronounceability. A name people mispronounce in a serious room costs you credibility every time.
• Competing on word of mouth — consumer, local, community — weight repeatability. The name has to survive being said across a table in a loud restaurant.
• Competing on search — marketplaces, tools, content — weight distinctiveness. A name you cannot rank for is a name you will pay for forever.
• Competing on enterprise sales — weight how it reads in a procurement document beside names people already trust.

What if none of the candidates feel right?

Then the brief was wrong, and generating another hundred names against the same brief will produce the same disappointment. This is the most common failure in naming, and it is almost never a creativity problem.

The fix is to write down what is wrong with each rejected name in the specific: too clinical, sounds like a law firm, too abstract, I know a company like this. Those rejections are worth more than the shortlist — at intake, nearly every founder writes some version of “bold, confident, premium, modern,” but shown forty real candidates, the same founder knows precisely what is wrong within seconds. That reaction is the brief they could not write. Rewrite the brief from it, and only then generate again.

The order, in one list

1. Cut on hard constraints — trademark, real prior use, spelling, language.
2. Shortlist three to five. Put the scores down.
3. Say each one out loud, on a call, in an email address, on an invoice.
4. Choose the category — descriptive, suggestive, coined — from your distribution.
5. Break the tie on the one dimension you actually compete on.
6. If nothing fits, fix the brief, not the name list.
7. Before you commit, clear it properly — and talk to a trademark attorney before you file.

Brand Clearedruns this end to end: five AI creative agencies generate roughly a hundred candidates, every one goes through the availability gauntlet — domains, the USPTO register scoped to your Nice classes, a common-law web sweep, the Apple App Store and Google Play, seven social platforms, famous-name collisions and historical use — and the surviving names are scored on seven brand and five growth dimensions, then argued over by the same four-panel audit whose disagreements this article is built on. About 10–15 minutes, $149. It is research, not legal advice.

You can also check any single name free — domain, app stores, seven social platforms and a famous-entity collision, in seconds, no signup.