The name usually gets chosen in a meeting.
A founder has five finalists in a shared doc. The team has lived with them for a week, and now it comes down to a show of hands. One name keeps drawing the votes. It says easily, nobody stumbles over it, the domain is available, and when the founder reads it aloud the room agrees it sounds right. So that is the one. The brand ships.
Two years later it sits on a shelf beside four competitors that ran the same meeting and reached for the same kind of name. A shopper scanning the category cannot tell them apart. The name was not a mistake. It cleared every check the founder knew to run. The problem is the checks. The process that produced it is the same process nearly every guide online will hand you, and it quietly skips the steps that would have made the name sharp instead of safe.
The same basic process works. It just has to include three steps the usual version drops.
The process every guide gives you
Read the guides that rank for naming a brand and the advice converges fast. Some lay out five steps, some eight or nine, but the arc rarely changes. Define your brand and your audience. Generate a long list of options. Cut it to a shortlist. Run a trademark search. Pick one and register it.
There is nothing wrong with that arc. It is a sensible scaffold for the visible work of naming, and it catches the obvious failures. It will stop you from picking a word that is already trademarked in your category, or one that means something unfortunate in another language.
What it leaves out sits underneath the visible work. Three steps decide whether the name you end up with is sharp or forgettable, and the standard process either rushes them or drops them. It never maps the real shelf you are about to compete on. It tells you to pick the option you like best, which is the single move you most want to avoid. The third gap is validation, treated as nothing more than a trademark search. Put those three back, and the rest of the process does what it was supposed to do.
The brand naming process, step by step
- Decide what you are naming, and at what altitude. Name the brand, not the single product in front of you.
- Map the shelf before you fall in love. Lay out the real names you will compete beside and see where they cluster.
- Generate wide, and keep it loose. Produce a lot of raw options before you grade any of them.
- Throw away the names that feel right. The obvious favourite is the one everyone lands on. Cut it on purpose.
- De-risk the finalist with evidence. Trademark, domain, a real customer read, a language check.
- Commit to a decision you can defend. Pick with a written reason, not a gut nod.
Steps two, four, and five are the ones the generic guides skip.
1. Decide what you are naming, and at what altitude
Most processes open here, and they are right to. Before any names, you write down what the brand is, who it is for, and how it should feel in a buyer's hands, whether that is clinical and exact or warm and a little reckless. That short brief is the ruler every later step measures against.
The piece the standard version gets wrong is altitude. Founders tend to name the product sitting in front of them, the one item they are about to launch. A name pinned that tightly to a single product becomes a cage the moment the line grows. You name a barrier cream and call it something that shouts cream, then you add a serum and a cleanser and the name is fighting you.
Name the brand one level up, at the altitude that still fits when there are three products under it, or ten. The brief you write should describe the company you are building, not the first thing it sells.
2. Map the shelf before you fall in love
This is the first step the guides skip, and the rest of the process leans on it.
Before you write a single name, build a map of the shelf you are about to join. Not a vague sense of your competitors, but the actual names, lined up, each one read for what it is doing. List the names you would see in one aisle or one category search, twenty or thirty of them, and group the ones making the same move. In any crowded category they cluster, because the obvious association is obvious to everyone, and one founder after another reaches for it.
Clinical skincare is the clearest example. Look at the shelf and a single move dominates. CeraVe builds its name from ceramides, the skin-barrier lipid it is formulated around. Cetaphil is a scientific-sounding coinage with the same intent, a name engineered to read like something a dermatologist would trust. Brand after brand borrows the language of the lab. That repeated move is the convergent zone, the spot on the shelf where everyone is already standing.
The open ground, the white space, is everywhere they are not. The Ordinary named itself the plainest thing imaginable and stood out by refusing the lab-coat vocabulary the whole category leaned on. It read the crowd and walked the other way.
You cannot find that open ground until you can see the crowd. The map is what makes the obvious answer visible as obvious, which is exactly what you need before the next step.
3. Generate wide, and keep it loose
Now you generate, and this is the part that got easy. A founder with a clear brief and any decent AI tool can produce hundreds of raw options in an afternoon. Generation is close to free.
So the urge to over-engineer it is worth resisting. It is tempting to feed the model your whole strategy, your positioning, your list of must-have traits, and ask it to thread all of them at once. That produces careful, on-brief, forgettable names, because a heavily instructed prompt walks straight to the obvious answer. A loose, almost naive prompt fed a few sharp example names will range further and turn up more surprises. Ask for a hundred options in the spirit of a few names you admire, then keep the twenty that make you stop.
Keep this step wide and unprecious. Some of what comes back will win on meaning, some purely on sound, the way a name can land before a buyer has parsed what it means. That second lever, the sound itself, is its own subject, taken up in the guide on sound symbolism in brand names. You are not grading yet. You are filling the pool.
4. Throw away the names that feel right
Now the part that feels wrong to do. The standard advice says to read your long list, find the names you like best, and shortlist those. Follow it and you will almost always shortlist the convergent names, the ones sitting in the middle of the crowd you mapped in step two.
They feel right precisely because they are familiar. Your brain has seen that shape of name a hundred times on that shelf, and familiarity reads as rightness. Every other founder in your category feels the same pull toward the same names. So does every AI generator, trained on the same crowded internet.
So invert the instinct. Treat the names that feel most obviously right as suspects, not finalists. The first job of the cut is to strip out those convergent names on purpose, the ones a free tool would have handed you anyway. What survives is the material worth taking seriously.
From there, cut toward the open ground you found on the map, and keep one name that makes the room slightly uncomfortable. The safe name, the one nobody objects to, is also the one nobody reacts to, and a name no one reacts to is a name no one repeats. Liquid Death named canned water after the one idea the whole beverage category works to avoid. Plenty of people hated the name on sight. The brand is now valued at $1.4 billion, built on far more than its name, but a name that polarizing plainly did not hold it back. The deeper case for why the safe option loses is its own argument, taken up in how to name a brand.
A good cut hurts a little. If every name you kept feels comfortable, you have probably kept the wrong ones.
Hundreds of raw names narrow to one finalist. Generation is cheap, and the cut is the work. Teal marks the three steps generic guides skip.
5. De-risk the finalist with evidence
Most processes stop one step too early, at the trademark search, as if clearance were the same as validation. Clearance only tells you the name is available. It says nothing about whether the name works. Real de-risking runs four checks, and a finalist should clear all of them before you commit.
Start with the legal screen, because it can kill a name outright. Run a free preliminary search on the US Patent and Trademark Office's Trademark Search tool at tmsearch.uspto.gov, and the WIPO Global Brand Database at branddb.wipo.int, plus the national office for any market that matters, for names you will use abroad. Those are knockout searches, not a legal opinion. Once a name survives them, a trademark attorney runs the full clearance before you spend a dollar on a logo.
Then check the practical availability. The domain, the social handles, and what comes up when you type the name into a search engine cold.
Then put the name in front of strangers, not your co-founder. Show a handful of real buyers the name with no pitch attached and ask what they think the brand sells and who it is for. If they place it in the wrong category, the name is fighting you. Mild curiosity is fine, since a strong name can intrigue before it explains. Active misdirection is the problem.
Last, run a language and culture screen if you sell across borders, so the name does not carry a meaning you never intended.
A name that clears all four is one you can stand behind.
A finalist clears four evidence checks before you commit. A trademark and legal screen, practical availability, a cold read from real buyers, and a language and culture check.
6. Commit to a decision you can defend
The last step is the one founders rush, because by now they are tired of the question and ready to move. Resist the urge to pick on a feeling.
Write down why the name won. The brief it serves, the open ground it claims on the shelf, the checks it cleared, the two runners-up and why they lost. A decision you can put in a paragraph is one you can defend later, to a co-founder, an investor, or yourself at 2am when the doubt arrives. A name chosen on a gut nod has nothing to fall back on, and the doubt always comes.
That written reason is the real output of the process. The name is just the part you print on the label.
Frequently asked questions
What are the different types of brand names?
Most names fall into a few types. Descriptive names say what the product is, like General Motors. Suggestive names hint at a benefit, like Airbus. Coined names are invented, like Kodak or Cetaphil. Associative names borrow a related idea, like Amazon for scale. Eponymous names use a founder, like Dyson. No type is better than another. What matters is whether the name fits the brand and stands apart from the rest of the shelf.
How long does it take to name a brand?
Faster than an agency, slower than it feels like it should. Run it yourself and the generating is a day, but the cutting and the validation are the slow parts, so protect a focused week or two. Naming agencies typically take three to six months. The real clock is trademark clearance, which runs on the attorney's and the office's schedule, not yours.
Do you need the matching .com domain?
Less than founders think. An exact-match .com is a nice-to-have, and chasing one is a common way to talk yourself into a worse name. Plenty of strong brands run on a modified domain or a newer extension and let the trademark, not the URL, carry the weight. Decide whether the name is clear and ownable first, then solve the domain.
Can AI name your brand for you?
It can do half the job. AI is good at the generating, producing a wide pool of raw options in minutes. It is poor at the deciding, because it converges on the same obvious names every other founder's prompt produces and has no read on the specific shelf you compete on. Use it to fill the pool, then do the cutting and the validation yourself, or with someone who knows the category.
BrandNames runs this process for founders, the wide generation, the hard cuts, and the evidence behind the finalist, with a strategist on every project. See how it works, or drop your email for one note when the doors open.