Skip to content
Business Naming

How to brainstorm business names that are not random

8 min read

Ask anything for business name ideas and you get a list of strangers: technically relevant words, arranged plausibly, none of which feel like a company. The output is not the problem — the request was.

Naming a business well is a process with stages, and the stages want opposite mindsets. Generating wants abundance and no judgement. Choosing wants strictness and no attachment. Almost every bad naming session is those two happening at the same time.

This is the whole sequence, from the brief through to the moment you stop looking — including the question people actually mean when they ask for something that sounds professional.

Start with a brief, not a blank page

Half a page, written before anything is generated: who the customer is, what the business does, the personality you want, where you operate, how long the name can comfortably be, and how far the business might travel from what it does today.

The brief does two jobs. It gives every idea a reason to exist, so a candidate can be rejected for a stated reason rather than because somebody in the room shrugged. And it becomes the criteria you score against at the end, which is what stops the final decision from being won by whoever is most stubborn.

Build territories before you build names

A territory is a direction rather than a word. Four or five of them, each a small cluster of related concepts: the outcome the customer gets, the emotion attached to it, a metaphor, the objects and places involved, the movement or change the business creates, and the sounds you want.

Territories are what stop a naming session from becoming a thesaurus exercise. If every candidate is a synonym for what you sell, you are competing for the same handful of stems as every competitor — and the fact that all of you had the same idea is why none of the domains are free.

They also give you somewhere to go when a whole direction turns out to be unavailable, which is a normal outcome rather than a crisis.

Choose seeds deliberately

From those territories, pick five strong seed words. A seed should represent something specific you want to preserve — a sound, an image, a meaning, a mood — and you should be able to say which in one sentence.

Five deliberate seeds beat fifty keywords. Fifty keywords produce fifty unrelated lists; five seeds produce five families of names that share a personality, and a family is something you can learn from.

Transform, do not restart

This is the mechanical core, and it is the step most people skip in favour of thinking of new words. Take a promising root and work it: rotate the vowels, change the ending, change the opening consonant, add or remove a syllable.

Controlled mutation explores deeply where new ideas explore randomly. The advantage is that you can tell what is working — when three of your favourites share an ending, that ending is a finding, and the next round should be built on it. Fifty new unrelated words teach you nothing about the fifty-first. Fifty ways to transform a word into a brand name is the catalogue of edits.

Use constraints as creative tools

Constraints feel like they reduce creativity and reliably improve it. Two or three syllables. No consonant cluster English does not use. One obvious pronunciation. No letter doing no audible work.

Each of those removes an enormous space of bad candidates without removing any good ones, and it makes the remaining output comparable. An unconstrained list is unsortable, which is why it feels like noise.

Separate generating from judging

Generate the whole raw set first. Do not check a domain after every idea, and do not evaluate as you go. Then stop, change mode, and filter hard.

Mixing the two lets availability bias your taste, and it does so invisibly. A name marked available feels better than the same name marked taken — so if you check as you generate, the checker chooses your brand, and it will hand you whatever nobody else wanted.

The rule that follows: an available domain confirms a candidate, it does not rescue one. If you would reject the name with the domain taken, reject it with the domain free.

Group into families, then delete most of it

Sort what you keep by seed and by sound rather than into one flat list. Comparing families usually reveals that you want something different from what the brief said — that the soft consonants are consistently winning, or that the abstract inventions are beating the descriptive roots.

Then delete aggressively. Most of what you generated should disappear: anything confusing, derivative, awkward to say, hard to spell, or carrying an unwanted meaning in a market you care about. Expect to keep something like one in ten, and treat that as the process working rather than as a bad session.

Make it sound established without pretending to be old

'Sound professional' is what people usually mean when they say a name feels random, and it is not about sounding like a century-old law firm. A credible name feels intentional. Five things produce that, and none of them involve gravitas.

Confidence instead of explanation

New founders try to fit the whole business into the name, which produces long generic combinations. A short distinctive name beside a clear tagline reads as more established, because the name and the explanation have separate jobs and each does its own properly.

Obvious pronunciation

A name people hesitate over starts every introduction with friction. Invented words are fine — the spelling just has to give a strong clue about the sound. A hesitation is not a one-off; it happens in every meeting for the life of the company.

No trend stacking

A fashionable suffix, an unusual spelling and a buzzword are each individually defensible. All three at once reads as generated rather than designed, and a name that announces its vintage is a name you will want to change. Use modern patterns one at a time.

It survives boring contexts

Put each finalist on an invoice, an app icon, an email address, a support reply, a job listing and a bank statement. A business name lives in unglamorous places far more than on a homepage, and some names that look excellent in a typeface are quietly ridiculous on a payment confirmation.

It is used consistently

The same core spelling on the domain, the profiles and the product. You will not get every handle, and that is survivable — what damages credibility is unnecessary variation, because a customer who sees two slightly different spellings wonders whether both accounts are really yours.

Leave the name room to grow

A highly specific name can sound impressively professional today and misleading in two years. The invoice tool that becomes an accounting platform, the city in the name when you open the second office.

The name does not need infinite flexibility. It should just not create an obvious ceiling if expansion is part of the plan — and whether it is worth trading some immediate clarity for that room is the subject of short versus descriptive business names.

Write down why each finalist works

One sentence per finalist, connecting the name to something in the brief. This is a better test than it sounds.

If you cannot explain why a name belongs to this business, it may still be random — it has simply become familiar to you through repetition, which feels identical from the inside. The sentence is also what you will say to everybody who asks about the name for the next five years, so it is worth finding out now whether it exists.

Then check what is already using it

Only now, and only on the survivors. Search the exact name and its near-spellings across the web, the company registers where you will trade, the app stores, and the trademark databases for your markets.

Credibility disappears fast if a new name is easily confused with an established company, and an available domain is not clearance for anything. How to check if a business name is already being used is the detailed version.

How to know you have finished

The right name rarely arrives with certainty. Waiting for a feeling of obvious perfection is a way of generating forever, because that feeling is not reliably produced by good names — it is produced by novelty, and it fades on the third day either way.

Use evidence instead of certainty. You are done when a candidate clears all of these and no alternative meaningfully beats it:

The stopping conditions

  • People who see it can say it without a tutorial — accent variation is fine, fundamental disagreement is not
  • People who hear it can spell it, or at least make the same recoverable mistake
  • It survives a delayed memory test against several distractors
  • It fits the personality you wrote in the brief
  • It does not create an obvious ceiling on plausible growth
  • The domain trade-off is one you understand and would still accept in a year
  • Nothing in the conflict research disqualifies it
  • Two more rounds of generation produced nothing better

That last one is the real signal. Once new candidates consistently lose to your current top three, further brainstorming is procrastination with a productive shape, and choosing is the part of the work you are avoiding.

In short

Brief, territories, seeds, controlled variation, then filter — with availability checked last so it cannot bias your taste. And stop when two rounds produce nothing better than your top three, rather than waiting for certainty that never arrives.

Frequently asked questions

Why do generated business names always sound random?

Because they were generated from a category rather than a seed. A tool given 'business name for a consultancy' has nothing specific to work from and returns the average of everything, and averages are indistinguishable from each other.

How many candidates should a naming session produce?

Enough that deleting most of them feels comfortable — a few hundred raw is normal. The working shortlist afterwards should be three to ten, small enough to compare carefully.

Is a made-up word a real business name?

Yes, and it has two practical advantages: it is far more likely to be available across the domain, the handles and the trademark class, and it is easier to protect. The trade is that the tagline and the homepage have to do more explaining at the start.

Should I pick the shortest candidate?

Not by default. Short helps only while the name stays readable, and a short name people have to ask about costs more attention than a longer one they read at a glance.

Sources

These support factual background only. The guide itself is original writing, and nothing here should be read as legal advice.

Related guides