Brand Naming (Placek Methodology)

This skill encodes David Placek's naming methodology (Lexicon Branding, ~4000 projects, 40+ years). Use it whenever the user wants to invent a brand-level name.

First step: pick a mode

Ask the user once: Quick or Full?

• Quick (~5 min): user gives a one-line brief, you return ~50 candidates grouped by direction with sound-symbolism + processing-fluency rationales. Skip the Diamond and disguised briefs.
• Full (~30–60 min interactive): run the Diamond exercise, generate against 3 disguised briefs, target 500–1000+ raw ideas, then screen down to a shortlist. Use when stakes are high (company name, public product launch, rebrand).

If they don't specify, recommend Quick for features/internal tools, Full for company/public product.

Core principles (apply in both modes)

These come from Placek directly. Internalize them — they drive every decision below.

1. Distinctiveness beats descriptiveness. "CloudPro" loses; "Azure" wins. Descriptive names disappear into a sea of competitors. A great name starts a story, not makes a statement.
2. You won't know it when you see it. The right name almost always feels uncomfortable on first read. Comfort = familiarity = past = forgettable. Don't filter on "do I like it" — filter on "what does it do".
3. Polarization is a signal of strength. If half the team loves it and half hates it (Pentium vs. ProChip), that's the right name. Consensus on a name usually means it's safe and weak.
4. Behavior and experience > positioning and values. Ask "how do you want to behave in the market, and how should the market behave toward you?" — not "what are your values."
5. Tangible-ize the intangible. Software/AI/services need a physical metaphor (Windsurf = flow, Blackberry = device-as-fruit). If the product is abstract, force a concrete image.
6. Compounds are multipliers. Powerbook, Facebook, Blackberry, Windsurf. Two familiar morphemes → 1+1=3 associations. Don't dismiss compounds for being "long."
7. Suspend judgment; speculate. Generate volume first. 200 names isn't enough — push to 1000+. Then ask "what could we do with this?" not "is this good?"
8. The .com doesn't matter much. Get the right name first; buy the domain second, or use a prefix/.ai/.co.

Mode A: Quick generator

When user picks quick mode:

1. Gather a one-paragraph brief: what does the product do, who's it for, how should it feel, what's the competition called.
2. Generate at least 50 candidates across these direction buckets (label each bucket):
   • Coined / invented (Vercel, Sonos, Pentium) — morpheme + morpheme
   • Compound (Windsurf, Blackberry, Powerbook) — two familiar words
   • Metaphor / experience word (Azure, Swiffer, Cursor) — evokes feeling
   • Mythological / classical (Anduril, Anthropic, Atlas) — use sparingly; warn this is the engineer-default trap
   • Cross-domain pull (the "magazine trick" — pick a word from a totally unrelated field like sailing/cooking/architecture)
3. For each name, write one line of rationale citing: sound-symbolism letter assets, morpheme meaning, processing fluency, and what the name does for the reader. See references/sound-symbolism.md.
4. Mark candidates with a ⚡ if they are likely to polarize the team (this is good — call it out).
5. Do NOT pre-rank by "best." Present neutrally so the user can react.
6. Close with: "Test the top 3 by telling a friend 'a new competitor just launched called [name]' and listen to what they imagine. Don't ask 'do you like it.'"

Mode B: Full workflow

When user picks full mode, run these phases in order. Each is interactive — pause for user input between phases.

Phase 1: Identify (the Diamond)

Walk the user through the Diamond exercise. Draw it in text:

              WIN
       (how do we define winning?)
             /   \
            /     \
   WHAT TO -+------+- WHAT WE
     SAY    |      |   HAVE
            \     /
             \   /
        WHAT WE NEED

Ask each corner in order:

• WIN — how do you define winning? (Five people in a room will give five answers. Surface them all.)
• WHAT WE HAVE — what do we already have that helps us win?
• WHAT WE NEED — what's missing? (Talent, distribution, capability, etc. People often say "a good name" — correct them: the right name, not a good one.)
• WHAT TO SAY — what do we need to say to the market to win? Spend the most time here.

Then the behavior question: "How do you want to behave in the marketplace, and how should the marketplace behave toward you?" Then the competitive landscape: list 5–10 competitor names and the language they all use.

Synthesize into a creative framework (3–5 bullets). This replaces "objectives" — it's a window, not a target. Get user sign-off before moving on.

Phase 2: Invent (disguised briefs)

This is the unique Placek move. Generate against three parallel briefs:

• Brief A — real brief. Generate ~200 candidates directly against the framework.
• Brief B — adjacent disguise. Pretend the product is in a different category (if it's an AI IDE, pretend it's a kitchen appliance, or a sailboat, or a sports brand). Generate ~200 candidates for that product.
• Brief C — pure cross-domain. Pick a totally unrelated context (hunting magazine, architecture journal, cooking show, a sport). Generate ~200 words/phrases from that world. Don't try to make them "fit" — just collect.

Most winning names come from B or C. Windsurf came from a "flow / dynamic movement" list, not from an "AI coding tool" list.

Use references/sound-symbolism.md to bias letter choice toward the framework's desired feeling (alive → V/Z, reliable → B/D, fast → X/K, calming → soft S/L/M).

Aim for 500–1000+ total ideas before any screening. Resist the urge to evaluate during generation.

Phase 3: Screen (the funnel)

Run candidates through these gates in order. Each gate eliminates a lot.

1. Trademark sanity — flag obvious conflicts (existing famous brands, identical strings in the same category). For real legal clearance the user needs a TM attorney; say so.
2. Linguistic / cultural — flag words that mean something bad in a major language, sound like a slur, or share a string with a known disaster/scandal. If unsure, say "verify with a native speaker of X."
3. Processing fluency — drop anything hard to pronounce, spell, or remember on first read. Vercel passes (ver + cel, both familiar). "Xqzlith" fails.
4. Distinctiveness vs. landscape — drop anything that sounds like an existing competitor or uses the category's tired vocabulary.

Output: a shortlist of 10–15 survivors, grouped by direction.

Phase 4: Implement (pressure-test the shortlist)

Don't ask "which do you like." Run these drills with the user:

• Polarization check. Show the list to 3–5 teammates. Names where opinion splits are good signals (Pentium story). Names everyone agrees on are usually weak.
• Competitor drill. For each top candidate, frame it as: "A new competitor just launched called [name]." Ask what the listener imagines. The name's job is to fire imagination.
• "They're not like the other guys" test. A winning name makes a stranger say "I don't know what they do exactly, but they're not like the others."
• Prototype it. Mock the name into 2–3 contexts: a logo treatment, a headline ("[Name] raises $50M Series B"), a product label. Does it carry?
• Future-flexibility. Will the name still work if the product pivots 15 degrees? Sonos worked because it was about sound, not "entertainment."

Recommend a top 3. For each, write a one-paragraph rationale the user can take to stakeholders — armor for the inevitable "I don't like it" reaction. Remind them: boldness > comfort. There is no power in comfort.

When to push back on the user

If the user gravitates toward a safe descriptive name ("CodeFlow", "DataPro", "AIAssist"), gently surface the Azure-vs-CloudPro lesson and ask them to live with a bolder candidate for 24 hours before deciding. Don't be preachy — just name the pattern.

If the user is renaming an existing product, confirm the trigger fits one of Placek's three valid reasons: (1) the original was a throwaway "we just needed something" name; (2) the company genuinely pivoted; (3) a merger or major capability expansion. If none apply, the rename may not be worth the cost.

Reference

references/sound-symbolism.md — letter-by-letter sonic and semantic associations Placek's team uses to bias name construction, plus compound/morpheme guidance and the named-examples library. Load when generating candidates or writing rationales.