Should I create a new category or join an existing one?

Default to join. If buyers already use a label when they describe your problem space, that label has distribution you didn't pay for. Create only when the evidence shows no coherent label exists or the existing one genuinely misdescribes you — and budget for the education either way.

What makes a category name stick?

Observably: buyers can repeat it to a colleague with the meaning intact, engines parse it close to the intended definition, and it doesn't collide with an established meaning elsewhere. Unprompted repetition by third parties is the signal that it's taking.

What's the cost of inventing a category name?

Until the market adopts your term, questions are asked in other words, and content filed under yours is absent from them. Creation means paying for two vocabularies for a while — publishing under today's words to be found while investing in tomorrow's.

How do I test a category name with AI engines?

Ask an engine cold what the term means. A name that parses sensibly gets explained approximately right before anyone has written about it; one that doesn't gets confidently misdefined. Re-run the probe after launch content ships and watch whether the parse moves toward your definition.

How to name a category

Guide · Founder · 5 min read · last verified 2026-07-27

Reviewed before publication Editorial board Independent commercial review

In shortA founder's method for category naming: default to joining categories buyers already use, argue creation from evidence, and put candidate names through the colleague, parse, and collision tests before money follows.

Naming a category is choosing the label a market will use to describe a class of products — including yours — when you are not in the room. That last clause is the whole discipline. A category name is not a possession; it is a word you are asking strangers to teach each other. You do not own it. You lease it from the market's willingness to repeat it, and the lease is revocable without notice.

Founders tend to approach the naming question as a creative exercise. It is closer to an evidence question with a creative step at the end, and getting the order right saves years.

The prior question: join or create

Before any name is invented, an earlier decision has to be made honestly: does a category already exist for what you do? Not "does a category exist that flatters us" — does one exist that buyers already use when they describe the problem you solve.

The default answer should be join. If buyers already reach for a label when they ask about your problem space — in forums, in procurement documents, in the questions they put to AI engines — then that label has distribution you did not pay for, and joining it means every question asked in its vocabulary can find you. Creating a category means declining that inheritance and funding a replacement from zero. Sometimes that is right: the existing label may genuinely misdescribe you, drag you into comparisons you lose, or bind you to a story in decline. But creation is the exception that must argue for itself with evidence, not the romantic default it has become in founder culture.

The test is checkable, not rhetorical. Gather the questions buyers ask about your space and look at the words in them. Ask engines cold what tools like yours are called and see whether a consistent label comes back. If a label exists and coheres, you are choosing whether to fight distribution. If nothing coherent comes back — different sources use different words, engines improvise — the ground is genuinely open, and creation stops being vanity and starts being an option.

Three tests for a candidate name

Whether joining with a twist or creating outright, a candidate name should survive three cheap trials before any money follows it.

The colleague test. Say the name once to someone who fits your buyer, then have them describe it to a colleague a day later. The name passes if the sentence "we're looking at one of those ___ tools" survives the retelling with the meaning roughly intact. Names fail this test by being too clever (the metaphor needs the founder present to work), too generic (the colleague hears a different product entirely), or too long (the retelling truncates it into something else).

The parse test. Ask an AI engine, with no context, what the term means. You are checking whether the words compose into your intended meaning on their own. A name that parses sensibly gets explained approximately right even before anyone has written about it; a name that does not will be confidently misdefined, and every generated misdefinition becomes a small piece of anti-marketing you did not commission. Run the same probe again after launch content ships — whether the parse moves toward your definition is one of the earliest signals that the name is taking, and the kind of change worth tracking against a locked baseline, which is the sort of watch a platform like Magrios keeps.

The collision test. Search the candidate honestly. Does it already mean something in an adjacent field, a programming language, a medical context, another market's jargon? Collisions are not always fatal, but a name that must out-shout an established meaning starts the education program mentioned above with a competitor it can never sign a truce with.

The cost of a name nobody searches

Here is the accounting most naming exercises skip. Every asset you publish — every page, comparison, guide, and answer — files itself under some vocabulary. If that vocabulary is a term nobody uses, you can execute brilliantly and rank first in a room with no audience. Questions are asked in the market's words; content filed under private words is simply absent from those questions, and no volume of production fixes a filing error.

This is the cost side of category creation stated concretely: until the market adopts your term, you are paying for two vocabularies — publishing under the words buyers use today so you can be found, while investing in the words you hope they use tomorrow. Companies that budget for both can afford creation. Companies that skip straight to the new vocabulary are not creating a category; they are hiding, expensively.

A name nobody searches is not always a mistake — every adopted term had a first day. It is only a mistake when nobody decided to pay for it.

Joining without disappearing

If the evidence says join, distinctiveness moves down a level. You take the category's name — that is the price of its distribution — and you differentiate through the approach, not the label. Name your method, your architecture, or your point of view within the shared category, and let that named idea do the work the invented category name would have done. This pattern keeps the inheritance (questions asked in the category's vocabulary reach you) while giving the market one new term to learn instead of two, and one is close to the limit of what a market will learn from a company it just met. The shape of the surrounding story — who defines the category, which criteria it carries — is its own subject, covered in What is a category narrative.

A name is a bet you keep paying

Whichever route you choose, the decision does not close at launch. The label's life shows up in observable places: whether buyers repeat it back unchanged, whether third parties use it without being asked, whether engines parse it closer to your definition this quarter than last. Watch those, on a schedule, and be willing to read bad news. Renaming later is costly — it resets corroboration you spent years accruing — but renting a dead word forever is costlier. The founders who handle this well hold the name lightly and the evidence tightly, which is the reverse of how it is usually done.