Create brand marks, badges, and identity systems via the Peters method — brand nouns, 50 sketches, noun combination, B/W presentation. Use when designing or evolving a logo or decoding 'make the logo bigger'. App icons and brand in iOS → apple-design. Triggers: logo design, brand mark, logomark, brand identity, badge design, rebrand, logo presentation.
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Um comando direto ignora o prompt de revisão. Verifique a origem antes de executá-lo.
Instruções da origem · Visualização somente leitura
name
logo-design
description
Create brand marks, badges, and identity systems via the Peters method — brand nouns, 50 sketches, noun combination, B/W presentation. Use when designing or evolving a logo or decoding 'make the logo bigger'. App icons and brand in iOS → apple-design. Triggers: logo design, brand mark, logomark, brand identity, badge design, rebrand, logo presentation.
A logo here is a containerless system — mark, logotype, badge, patterns, icons — that must work on a water tower, a favicon, and an embroidered hat. (An app icon is a single artwork in an OS-owned container — a different task shape; see Relationships.)
1. The ten pillars — 3 you can't control, 7 you can
The client-conversation tool for "why doesn't this feel iconic yet": "you cannot control the timeline, you cannot control the marketing budget, and you cannot control the product quality." The Swoosh wasn't iconic in 1971; iconic = repeat viewings written to long-term memory, which time and media spend buy, not design. Use the uncontrollables to set expectations at kickoff and to depersonalize the "it doesn't feel like Nike yet" complaint.
You can control: personal passion, universal beauty, originality, functionality, color, memorability (story), simplicity. End state: "When a mark is embedded in enough of your audience's long-term memories, you will not need to use words anymore."
2. The Brand Noun Process
Decide the subject matter before anyone draws. Before the noun list, run Shir's four-lane intake: business (product, problem, model, competitors), audience (pain, traits, where they'll see the mark), brand (name, mission, values, characteristics), and client taste (expectations, liked/disliked colors and shapes). A logo is "like a handshake between a business and its audience"; don't skip either side.
Designer lists 15–25 nouns (person/place/thing) that could be the mark's subject. Read the client's site and values first.
Non-nouns must convert: "you need to come up with a noun that serves as a metaphor for that word." Trust → image-search "trust icons" → shield, key, keyhole, shaking hands, lock. Abstract values never enter the list raw.
Split notes into three keyword lanes (Shir): terminological keywords (industry/product words), characterizing keywords (adjectives/traits), and figurative keywords (objects/symbols). The noun list comes mostly from the third lane, but every sketch should still map back to at least one keyword.
Homework: every client-side approver of the final logo makes their own noun list (teach what a noun is — most client lists come back less than half nouns).
Merge, organize by category, dedupe, then narrow to ~15 WITH the client in a meeting.
The payoff is procedural: "Narrowing down the list is more important than making the list. It eliminates everything that the client does not want to see. That way when you are selling logo ideas in the later stages of the project, the decision will only be based on the aesthetics of the design and not the subject matter." Subject-matter fights are settled before sketching; "Including the client in this process is absolutely paramount."
Guard the narrowing: clients will cull clichés (shield, apple, heart) not realizing combination makes them fresh (§4) — "You do not want to lose amazing subject matter because your client feels you will not be able to transform it into something original."
3. Sketch discipline
Minimum 50 rough sketches from the approved nouns. "Quantity is king." Rough = just enough to capture the idea; refining a sketch at this stage is wasted time ("You can't polish a turd" — generate, don't polish).
Time-box the warmup (Shir): don't spend more than 30 seconds on early sketches; keep them on the canvas instead of deleting/refining them. His lower floor is 20 sketches, but Peters' 50 remains this skill's standard.
Vectorize the best 15 with geometry, simple shapes, a grid. Thick line weight for small sizes, but "The negative space in a brand mark is just as important as the positive space." Try variants, always at small sizes, black-on-white and white-on-black.
Cull and refill: before the deck ships, identify the five weakest marks and go back to the sketchbook to replace them with something better.
4. Five noun-combination techniques
One noun is a cliché; two combined are ownable (shield = trite, lightning bolt = trite, shield+bolt = unique). The sketch-phase engine:
Overlapping geometry — sketch many simple versions of each noun, draw each one's geometric skeleton, hunt for shared curves (a droplet's curve doubling as a leaf vein). "The best combinations feel like they belong together. You don't have to work hard to combine them."
Negative-space combination — Firecraft Pizza: the flame's negative space defines the pizza slice's edge. Same hunt: pages of sketches per noun, scan for shape overlap. True visual harmony "is quite rare" — keep casting.
Visual rhythm — evenly spaced linework ties two nouns and adds energy (cross + shield). Overdo-warning: shave lines while the idea survives; "you are making a logo, not an illustration."
Visual flow — linework converging the eye to the mark's center (Shoreview's waves→trees→water loop). Renaissance compositional movement: "if this technique has worked successfully since the 1300s, it is a solid design principle." Favor nouns with built-in motion (arrows, chevrons).
The unexpected twist — cliché + cliché = ownable (apple + heart). The defense for keeping "boring" nouns on the list.
Warnings (Peters, verbatim or near):
"There is nothing more forgettable than a letter in a circle."
Combining client-picked concepts — "'Frankensteining Ideas,' as it's called in the design business, is usually a horrible idea." (The Kwikbit exception: blend the essence of one sketch with the outer shape of another, not parts onto parts — see references.)
Funny logos age like jokes: "they are only funny the first time you hear them." Fine for T-shirts; let humor live in copy (Epilogg).
Never pure black or white "unless [you have] a good reason" — warm or cool them to harmonize the palette.
5. Tests
The hat test (universal beauty): "mock up a potential brand mark on a hat. I then ask the client if they would be proud to wear this hat." Clever-but-unwearable fails.
Functionality floor: works as a 16×16 favicon; embroiders small; stamps into concrete if the client needs it. Embroider real samples when budget allows (Eagan staff uniforms).
Small-size failure diagnosis: find where it breaks first — "Overly thin lines and tight negative space tend to be the major issues I see in logo designs in small-scale applications." Fixes in order: thicker lines → more negative space between elements → simplify without losing the meaning.
Shir's extra checks: let it rest overnight before final judgment; test as a profile image inside a bounding box for optical centering; place it in different applications; run a quick visual-similarity/image-search check for uniqueness before presentation.
6. The presentation protocol (first round)
15 marks, black and white only — no typography, no color. Aesthetic-only decisions, remember, because nouns are pre-approved.
Open with a recap of the approved brand nouns; list each mark's referenced nouns on its slide (the PDF must self-explain later).
Each mark black-on-white and white-on-black; final slide = all 15 side by side.
Let the logo speak first (Shir): after context, show the mark alone with generous white space before explaining symbolism. Then explain what each element represents, show scale/color behavior, and use mockups so the client sees the mark "in action."
Lead: pick your top three in advance, place them for impact, and state the meeting's goal up front (client picks their top three). Frame feedback positively — say what you do like.
Never email the PDF. Present live to "answer questions and fend off bad feedback on the fly" — the most crucial presentation of the project.
The silent ballot (anti-HiPPO): with a large stakeholder room, present all 15 without commentary, then at the side-by-side slide ask everyone to "quietly write down on a piece of paper which three they would pick," then go around the room tallying aloud. "This keeps the person in charge on the client end from dominating the conversation." Start the reveal with your own top three.
7. "Make the logo bigger" — the reframe
"In most cases the size of the logo is not the real issue. The real issue is that the branding in the layout is not strong enough." (Earlier, sharper: when a client says it, "the problem is that the work is not on-brand.") Enlarging makes it worse.
Fix: integrate the brand mark's geometric DNA into every element — patterns, icon sets, badge shapes, photo crops, custom type, flourishes. Test with the PDCo squirrel standard: an on-brand icon shares the mark's line width, negative-space ratios, and curves, so seeing the icon alone "would make you think of the brand mark, therefore reducing the need for a gigantic logo on each page." Peters, after adopting this: "I have not been asked to 'make the logo bigger.'"
8. Badge design tier
A brand system is modular: mark (avatars/small), horizontal (compact), vertical (large), and badge for large storytelling surfaces — murals, packaging, posters. Don't force it: "for many tech companies a badge would feel out of place." Include one only if it strengthens the system.
Content hierarchy: "The company name should be the most important element, followed by the brand mark, then the tagline, and lastly the balance words." 15+ words is fine if balanced.
Balance words — visually equal-length pairs flanking the badge: "ESTD"/"2023", "TRD"/"MRK", "MPLS"/"MINN", "QLTY"/"CRFT". Vary them per brand or they stop telling a story.
Shape containers: circle/oval is the neutral default; better, derive the external shape and inner containers from the brand mark's geometry (the Wits roofline badge). Explore several; symmetrical and balanced wins.
Visual balance: every element on the left balanced by one on the right — and balance the negative space too. Can't balance it? Add elements; you make the rules.
9. Brand evolution
"When you start from scratch, you erase all of the visual brand equity. Even ugly logos can become memorable over time." Default to evolution versus revolution.
Run the audit that opens every PDCo redesign — two lists, presented to the client:
"Here's what was not working with the original logo:" (gradients, kerning, thin linework, illegible type, too many colors…)
"Here's what was working:" (the core idea, an iconic shape, a defensible color) — this is the equity you keep.
Order of attack: functional pain points first (ask the client what breaks in reproduction), then the concept (simpler way to say the same idea?), then color (fewer, more intentional). Worked audits (Urbane, Eagan Police, Rush Creek) in references.
"When a small pool of modern-day designers only find inspiration in each other, all of their work starts to look the same." Peters' practice: dig antique stores, estate sales, small-town museums (small towns = less picked over; regions differ) for pre-Rand/Bass "wild west" design — "lost trends and solutions to problems." Hunt three things: unique badges, bold brand marks, custom typography. Photograph, don't hoard; study, never copy. Prescription for any team whose work is converging on Dribbble-sameness: deepen the inspiration pool beyond living designers.
Checklist
Uncontrollables (time, budget, product) named to the client at kickoff?
Business/audience/brand/client-taste intake done; abstract values converted to noun metaphors; ~15 nouns narrowed with every approver?
Terminological/characterizing/figurative keyword lanes captured; every surviving sketch maps to a keyword?
50+ rough sketches before any vectorizing; early sketches time-boxed; combinations attempted via all five techniques?
15 vectorized; five weakest culled and replaced from the sketchbook?
No letter-in-a-circle, no Frankenstein merges, no joke marks, no pure black/white?
Passes the hat test, 16×16 favicon, profile-image box, real applications, embroidery; thin lines / tight negative space diagnosed?
Rested overnight and uniqueness/image-search check run?
First round presented live: B/W only, nouns recapped, logo shown alone before explanation, silent ballot if the room is big?
"Make the logo bigger" answered with brand-mark DNA in patterns/icons/type, not scale?
Badge (if any) earns its place; hierarchy name > mark > tagline > balance words?
The method assumes a human designer with a client in the room. Mapping for AI-agent execution (the process and thresholds are unchanged — only the media differ):
Sketch round = a batch of low-fi directions: one-line concept descriptions and/or rough SVG outlines, one idea each. Quantity still beats polish — keep the 50-idea bar, time-box each, and don't refine at this stage.
Vectorize the best 15 = clean SVG constructions from simple geometry; render each at small sizes, black-on-white and white-on-black, before showing anything.
Presentation / silent ballot = present ~15 candidates unranked, each with its referenced nouns listed; ask the user to pick their top three before revealing yours.
Hat / favicon / embroidery tests = small-size render checks: rasterize at 16×16 and thumbnail sizes and inspect for the §5 failure modes (thin lines, tight negative space), fixing in the §5 order.
Noun intake and narrowing still happen with the user — never skip to sketching from a self-generated noun list; the narrowing meeting becomes a message exchange, but the client-approval gate stays.
Relationship to other skills
apple-design (app-icon-design) — shared core (metaphor selection, simplicity, small-size testing: Peters' thin-lines/negative-space diagnosis = the squint/blur test family). Different task shape: an icon is a single artwork inside an OS-owned container and rendering system; a logo is a containerless multi-format system. Route icon work there.
ux-writing (naming-features-and-labels) — the noun-metaphor move (abstract value → concrete noun) is the visual twin of its candidate-generation exercises; its criteria judge the brand name this skill draws marks for.
client-work (client-engagements) — the presentation protocol and silent ballot here are logo-specific instances of its feedback system; qualification, kickoff, and failure handling for the engagement live there.
apple-design (ios-brand-identity) — once the identity exists, expressing it natively in an app (tint, type, components) is that skill's job.
client-work (pricing-creative-work) — the six-week scoped process is the thing being priced; pricing itself lives there.
Staleness note
The method is durable — noun-driven subject matter, quantity-first sketching, combination techniques, equity-preserving evolution are tool-independent. Decaying specifics: named cases and tools are 2023-era (Adobe Illustrator workflow, Behance/Dribbble/Google image search as search surfaces, specific client examples); AI image search and generation change the mechanics of noun-metaphor hunting and originality checks, not the process.
