Naming Brainstormer
Naming fails in predictable ways: the team generates twelve options in one style, falls in love early, and discovers the problems after the announcement. This skill forces width before depth — candidates across several distinct naming territories — then applies unsentimental filters and a scoring pass, so the shortlist survives contact with real mouths, real spelling, and real markets.
When to use this skill
- Naming a product, feature, company, internal tool, or open-source project
- Renaming something whose current name collides, confuses, or embarrasses
- The team has a favorite and needs it stress-tested against honest alternatives
- A launch deadline demands a defensible shortlist this week, not a perfect name someday
Workflow
- Write the naming brief first. One line on what is being named, who says the name aloud and in what sentence ("did you check it in ___?"), the feeling it should carry, names it must not resemble, and hard constraints: length, language markets, works as a command, works as a handle.
- Generate across territories, not in one pile. Produce eight to twelve candidates in each of
at least five territories:
- Descriptive — says what it does
- Suggestive — evokes the benefit without naming it
- Metaphor — borrows an image from another domain
- Compound — two words fused into one
- Coined — an invented word with the right sound
- Human or place — a name-like word borrowed from an unrelated context
- Kill fast with mechanical filters. Cut anything that cannot be spelled after being heard once, cannot be pronounced after being read once, runs past four syllables, carries an unfortunate meaning in a target market's language, or collides with a well-known thing nearby.
- Run the sentence test on the survivors. Say each name inside the user's actual sentences: "___ is down", "just paste it into ___". A name that feels wrong in the sentence is wrong, whatever it looks like on a slide.
- Score the shortlist with the rubric below. Carry the top five to ten forward with one line of rationale each, and force at least two finalists from territories other than the team's early favorite.
- Hand off the human checks explicitly. Trademark searches, domain and handle availability, and native-speaker review in every target market are required before commitment — and nothing in this exercise established legal availability.
Scoring rubric
Score one to five per dimension; weight distinctiveness double in a crowded category.
| Dimension | A five looks like |
|---|---|
| Clarity | A stranger guesses roughly what it is |
| Distinctiveness | Nothing nearby sounds or looks like it |
| Say-ability | Survives being said aloud in a meeting, twice |
| Spell-ability | Heard once, typed correctly |
| Stretch | Still fits if the product doubles in scope |
| Risk | No collisions or bad meanings found so far |
Guardrails
- Never declare a name available; availability is a legal and market question this skill cannot settle
- Do not let descriptive names win by default — they score high on clarity and age the worst
- Ten mediocre candidates in one territory are worth less than three each in five territories
- If every top scorer comes from one territory, the generation phase failed; go wider before deeper
- Present the shortlist without announcing your own favorite first; anchoring kills honest reads