business-name-fit – Marketing & SEO skill for AI agents

Business Name Fit

Overview

Help founders choose or check a business/brand/product name that is authentic to their origin and lands well in the market where they'll operate.

The core value is catching the mismatch between the two: a name can be perfectly good in its home language yet confusing, funny, or off-putting elsewhere. Examples:

• The Persian name "Anali" is fine in Iran but reads oddly to English speakers.
• The Swedish word "framåt" ("forward") can sound like "frame" to an English ear.

This skill both suggests new names and checks names the person already has.

Step 1 — Gather the essentials

If the skill is invoked with nothing else to go on (e.g. just /business-name-fit, no name, no context), don't launch into the full list below. Ask one question first, and wait for the answer:

> Are you looking to find a new name, or check one you already have?

Once that's answered, ask only for what's still missing from the conversation. Keep it to a couple of short questions.

1. Origin — the founder's culture/language or the company's home base (e.g. Persian/Iran, Swedish/Sweden, Mandarin/China).

2. Target market(s) — where the business will operate (e.g. international/English-speaking, EU, China, Arabic-speaking countries). There can be more than one.

3. Business area — the industry or field, because tone matters (a law firm and a candy brand need different feels).

4. Mode — do they want new name ideas, a check of names they already have, or both? (Already answered above if the skill opened with the find-or-check question.)

5. Existing names — if they're checking, collect the candidates.

6. Desired feel (optional) — modern, traditional, playful, premium, etc.

Step 2 — Run the fit checks (the heart of the skill)

Run every candidate name through these checks, once per target market:

• Meaning — Does it mean something unintended, negative, funny, or taboo in the market's language(s)? Include slang.
• Look-alike — Does it resemble an existing word that changes the impression (e.g. framåt → "frame")?
• Pronunciation — Can people in the target market say it easily, or does it get mangled?
• Spelling — After hearing it, can they spell it? Watch for tricky letter combos and origin-language diacritics (å, ø, ç, etc.) that won't survive in English.
• Distinctiveness — Where does it sit on the WIPO scale: generic, descriptive, deceptive, suggestive, arbitrary, or coined? Generic and descriptive names are weak and often cannot be registered at all — say so plainly, because a founder can build a business on such a name and still never own it. Reject deceptive names outright. Judge this separately per market: a word that is plainly descriptive at home may be arbitrary and strong abroad, and the reverse.
• Sound — Say it aloud. Does its sound-feel match the field? Front vowels (ee, i) read light and quick; back vowels (o, u) solid and heavy; hard plosives (p, t, k, b, d, g) sharp and firm; soft fricatives and nasals (s, f, m, l, n) smooth and gentle. Note honestly that these associations come mainly from English-language research and do not transfer automatically to other languages.
• Tone fit — Does it feel trustworthy and appropriate for that industry and that market?
• Origin authenticity — Does it still genuinely reflect the founder's origin, or has it been flattened into something generic?

Also weigh the classic evaluation criteria: relevance to the category, connotations, overall liking, ease of recognition, distinctiveness, and ease of recall.

references/naming-research.md holds the sources and reasoning behind these checks — read it when a founder asks *why*, or when a name category or sound effect needs explaining in depth. references/worked-examples.md shows three full cases end to end (a name fixed, a name approved, a name rejected) — read it when you need a model for how a finished analysis should read.

Be honest about confidence. If unsure whether a name carries an odd meaning or slang sense in a language, say so plainly and recommend a native-speaker check before the founder commits.

Step 3 — If suggesting new names

Every suggestion must satisfy three constraints at once: rooted in the origin, professional for the industry, and clean in the target market. A name that meets only two of the three is not a valid suggestion — drop it and find another.

3a. Write the naming brief first — before generating anything.

Founders (and firms generally) tend to work out what they actually want from a name only *after* they have fallen for a candidate, which corrupts the judgement. So write it down first, in one or two lines: what the business promises its customers, what feeling that promise needs, and what the name must therefore do. Then pick origin words that carry that meaning. Examples of the logic:

• Childcare → warmth, safety, gentleness. Not power or speed.
• Consultancy, law, finance → competence, stability, discretion. Not cuteness.
• Art authentication, certification, security → authenticity, precision, trust.
• Health → care, cleanliness, calm.
• Technology → clarity, motion, forwardness.

3b. Draw the raw material from the origin language.

• Real words, roots, names, places, or concepts from the origin language.
• Meaningful cultural ideas (nature, values, mythology) shaped into short, sayable forms.
• Light blends or coined words that keep an origin flavor.

Prefer a word whose literal meaning is itself the value proposition — a Persian word meaning "authentic" for an authentication company beats a merely beautiful word. But stop short of plainly describing the product: aim for names that *hint* (suggestive), use an unrelated real word (arbitrary), or are invented (coined). These are both more memorable and far more likely to be registrable than a descriptive name.

3c. Apply the professional-quality bar. Reject a candidate if it:

• is hard to say or spell in the target market after one hearing;
• carries a tone that clashes with the field (playful for a law firm, clinical for a toy brand);
• needs an accent or non-Latin character to read correctly;
• is longer than about three syllables, or looks like a random invented string;
• sounds like a personal first name when the business needs institutional credibility.

3d. Sanity-check against the market's naming conventions. Names carry different weight by region — what reads as confident in the Gulf may read as overblown in the Nordics. Make sure the name would not look out of place next to established firms in that field and that market.

3e. Run every surviving candidate through the Step 2 checks, then hand off to Step 4 to present them.

Offer 3–5 strong candidates rather than a long weak list.

Step 4 — Present the results

Default to a compact table, every time — checking existing names and presenting new suggestions both work this way. The whole answer should be readable at a glance: no scrolling through prose to find the verdict.

• Checking names — one row per name (one row per name × market if there's more than one market). Columns for whatever checks actually mattered for that name — not all eight every time. Cells hold a symbol plus 2–4 words (e.g. ❌ reads as slur (EN), ⚠️ crowded namespace), never a sentence.
• Suggesting names — one row per candidate: origin meaning, target-market read, verdict. Same rule — phrases in cells, not paragraphs.

After the table, add at most one short line naming the 2–3 strongest options. Never present a single favourite as if it were the only option.

Stop there. Do not add check-by-check breakdowns, reasoning paragraphs, or a written verdict for each name unless the person asks for more — "why", "explain", "tell me more about X", "detailed report" — in which case expand only the part they asked about, still as briefly as clarity allows.

Step 5 — Hand over the verification steps

Close every session with a short list of what still has to be checked — one line each, this skill cannot confirm any of it:

• Trademark — search the register in each target market before spending on the name (a formal step, not an afterthought).
• Domain & handles — availability where the founder will actually use them.
• Business registry — the company register in the home country.
• Native-speaker gut check — a real speaker in each target market, for the finalists.

Expand any of these only if asked.

Anti-Patterns

• Don't split the checks by feel alone. A name can pass the sound check and fail the look-alike or spelling check (or the reverse) — always run every check, don't stop once one check feels conclusive. See Scenario A in references/worked-examples.md.
• Don't call a name "safe" on meaning alone. Distinctiveness (the WIPO scale) is a separate, legal question. A name can be linguistically clean and still be generic or descriptive — weak and hard to register

… (full content is in the download package)