Branding
How to prompt a chat assistant (ChatGPT, Claude, Gemini) and image generators (Midjourney, Flux, Ideogram, Adobe Firefly) for brand strategy, naming, voice and visual identity when you are not a brand designer. Read fundamentals first for the general method; campaigns and copy are in marketing, print layouts in posters and print, screen design in UI/UX, and the science of palettes in color theory.
How experts think
Marty Neumeier's definition in The Brand Gap is the one most designers use: "A brand is a person's gut feeling about a product, service, or company." You don't own the brand; your customers hold it. What you control is the identity: the name, words, visuals and behavior that shape that gut feeling.
| Expert habit | What it means | What it changes in your prompt |
|---|---|---|
| Strategy before pixels | positioning, audience and personality are decided before anyone draws a logo | ask for the strategy first; feed it into every visual prompt |
| Different beats better | a brand wins by being distinct and memorable in its category, not by claiming "quality" | ask "what do all competitors say?" and then for the opposite |
| A logo is not a brand | the logo is one asset in a system of color, type, imagery, voice and behavior | ask for a system with rules, not one pretty mark |
| Consistency compounds | the same assets used the same way for years build recognition | ask for rules and examples of misuse, not just the ideal |
| Design for the worst case | a mark must work at 16 px, in one color, embroidered, on a busy photo | describe the hardest uses up front |
| Ownable, not just likable | a name or mark you can't register or that confuses buyers with another firm is a liability | screen names and marks legally before you fall in love |
| Test with the audience | stakeholders' taste is not evidence; target customers' reactions are | ask the model for test plans, not just opinions |
research ─► strategy ─► verbal identity ─► visual identity ─► guidelines ─► rollout
audit, positioning, name, voice, logo system, rules, touchpoints,
interviews personality messaging color, type, examples, launch,
imagery misuse measureVocabulary
| Term | What it means | Say it like |
|---|---|---|
| Brand | the reputation and feeling people associate with a company or product | "Describe the brand as a customer would after one year of use." |
| Brand identity | the designed, controllable expression: name, logo, color, type, voice | "Propose a brand identity system, not just a logo." |
| Brand strategy | the plan for who the brand serves, how it differs and how it should be perceived | "Write the brand strategy before any visuals." |
| Purpose | why the organization exists beyond making money | "Draft three purpose statements under 15 words; avoid clichés like 'empower'." |
| Mission / vision | what you do now / the future you're working toward | "Separate the mission (now) from the vision (10 years out)." |
| Values | principles that guide behavior; useful only if they rule something out | "For each value, give a decision it would force us to make." |
| Positioning | the place you want to occupy in a customer's mind relative to alternatives | "Position us against spreadsheets, not against other apps." |
| Positioning statement | an internal sentence: for whom, which category, what benefit, unlike what | "Write a positioning statement in Geoffrey Moore's format." |
| Target audience | the specific people you are for (and implicitly not for) | "Our audience is independent bookkeepers, not enterprises." |
| Category / frame of reference | the market you say you compete in; it sets buyers' expectations | "Suggest three category framings and what each implies about price." |
| Competitive alternatives | what customers would use if you didn't exist, including doing nothing | "List the competitive alternatives, including manual workarounds." |
| Point of difference | an attribute you have that alternatives lack and customers value | "Which of these is a real point of difference, and which is table stakes?" |
| Point of parity | something you must match just to be considered | "List the points of parity we must not neglect." |
| Competitive map | a 2-axis chart (perceptual map) placing competitors on attributes buyers care about | "Plot these 8 brands on a competitive map: price vs. playful–serious." |
| Brand promise | the one experience customers can count on every time | "State the brand promise in one sentence a support agent could keep." |
| Reason to believe (RTB) | proof that makes a claim credible: data, features, history | "Give two reasons to believe for each claim." |
| Brand personality | human traits the brand expresses, usually 3–5 adjectives | "Pick 4 personality traits and their 'but not' counterparts." |
| Archetype | a stock character (Hero, Sage, Jester…) used as shorthand for personality | "Which archetype fits, and which would be a cliché in our category?" |
| Voice | the consistent personality in words; it doesn't change | "Define our voice with 'this, not that' pairs." |
| Tone | how the voice adapts to the situation (an error message vs. a launch) | "Show the same message in our tone for a celebration and an outage." |
| Messaging hierarchy | ordered set of messages: tagline, value proposition, pillars, proof | "Build a messaging hierarchy from this positioning." |
| Tagline | a short public line that sums up the promise | "Give 15 tagline options, grouped by approach." |
| Boilerplate | the standard "About us" paragraph used in press releases | "Write a 60-word boilerplate in our voice." |
| Naming brief | the requirements a name must meet: meaning, tone, length, markets | "Here is the naming brief; propose names that meet every criterion." |
| Trademark | a legally protected sign (name, logo) identifying the source of goods or services | "Flag names that look likely to conflict; I'll run the legal check." |
| Wordmark (logotype) | the name set in distinctive type, no symbol | "Explore wordmarks with custom letter details." |
| Lettermark (monogram) | initials only | "A lettermark for 'NW' that works as an app icon." |
| Symbol / brandmark | a standalone icon: pictorial (a recognizable thing) or abstract (a shape) | "Abstract symbols only; no literal coffee cups." |
| Combination mark | a symbol and a wordmark together | "A combination mark with a stacked and a horizontal version." |
| Emblem | text enclosed inside a shape, like a badge or seal | "An emblem for merchandise; a simplified mark for digital." |
| Lockup | a fixed arrangement of logo elements (and sometimes a tagline or sub-brand) | "Define horizontal, stacked and symbol-only lockups." |
| Clear space / minimum size | the empty margin required around a logo / the smallest it may appear | "Specify clear space as a multiple of the symbol's height." |
| Palette roles | primary, secondary, accent, neutral and semantic colors, each with a job | "Assign each color a role and a rough usage share." |
| Type pairing | the combination of typefaces (e.g. a display face for headlines, a text face for body) | "Pair a characterful headline face with a neutral text face." |
| Imagery style / art direction | rules for photos and illustration: subject, light, crop, color treatment | "Write art direction for photography: subjects, light, framing, what to avoid." |
| Iconography | the family of small symbols in a product or on signage | "Describe an icon style: stroke weight, corner radius, grid." |
| Distinctive assets | elements people recognize without the name: a color, shape, sound, character | "Which of our assets could become distinctive if we used them for years?" |
| Brand guidelines | the rulebook: strategy, logo use, color, type, imagery, voice, examples | "Outline brand guidelines for a 5-person team without a designer." |
| Brand architecture | how a parent brand and its products or sub-brands relate | "Should the new product be a sub-brand or stand alone?" |
| Rebrand vs. refresh | changing strategy and identity / updating the look while keeping the equity | "Is this a refresh or a rebrand? Argue both, then decide." |
| Brand equity | the value (recognition, trust, preference) the brand has built up | "What equity would we lose if we changed the name?" |
| Touchpoint | any place a person meets the brand: site, invoice, packaging, support call | "List every touchpoint and what must change on each." |
| Mood board | a collage of references that sets a visual direction before design | "Describe three mood board directions in words I can give an image tool." |
| Vector vs. raster | shapes defined by math, scalable forever (SVG, EPS, PDF) / a grid of pixels (PNG, JPG) | "This is a raster sketch; list what a designer must fix when redrawing it as vector." |
Brand strategy
Positioning
Positioning is the decision everything else expresses. Two well-known formats:
Geoffrey Moore, Crossing the Chasm (positioning statement)
For [target customers] who are dissatisfied with [current alternative],
our product is a [new product category] that provides [key benefit].
Unlike [the product alternative], we have [key differentiators].
April Dunford, Obviously Awesome (five components, worked in this order)
1. Competitive alternatives what customers would do if we didn't exist
2. Unique attributes what we have that the alternatives lack
3. Value (and proof) what those attributes enable for customers
4. Target market who cares a lot about that value
5. Market category the context that makes the value obvious
(+ relevant trends, optional)Dunford's point is the order: start from alternatives, not from the category you wish you were in. Ask the model to work through the five steps with you, one at a time, rather than to "write our positioning".
Personality and archetypes
Personality is usually 3–5 traits, each with a "but not" to stop it sliding into a cliché: confident, but not arrogant; warm, but not cute. Archetypes come from Margaret Mark and Carol Pearson's The Hero and the Outlaw (2001), which adapted Jungian ideas into twelve brand characters. Treat them as a vocabulary for briefing, not as science: they are a framework from practice, not a validated model of how buyers behave.
| Archetype | Core desire | Typical voice | Risk in use |
|---|---|---|---|
| Innocent | safety, simplicity | optimistic, plain | bland |
| Explorer | freedom, discovery | adventurous, independent | generic "adventure" imagery |
| Sage | understanding | expert, measured | lecturing |
| Hero | mastery, winning | bold, motivating | chest-beating |
| Outlaw | disruption | rebellious, blunt | edgy for its own sake |
| Magician | transformation | visionary, mysterious | vague promises |
| Everyman (Regular Guy/Gal) | belonging | friendly, down-to-earth | forgettable |
| Lover | intimacy, pleasure | sensual, warm | over-promising |
| Jester | fun | playful, witty | jokes in serious moments |
| Caregiver | protecting others | nurturing, reassuring | patronizing |
| Creator | making something new | imaginative, crafted | self-indulgent |
| Ruler | control, stability | authoritative, polished | cold |
Ask the model which archetype every competitor already uses. If they are all Sages, a Sage brand disappears.
Voice and tone
Voice stays constant; tone flexes with the situation. Nielsen Norman Group describes four tone dimensions that make a voice concrete: funny vs. serious, formal vs. casual, respectful vs. irreverent, enthusiastic vs. matter-of-fact. Place the brand on each, then write "this, not that" pairs with examples.
| Trait | We are | We are not | Sounds like | Doesn't sound like |
|---|---|---|---|---|
| Direct | clear, short | curt | "Your invoice is overdue. Pay in one tap." | "We noticed that perhaps your invoice may be…" |
| Warm | human | gushing | "Nice work: that's your first month closed." | "OMG you're amazing!!!" |
| Expert | specific | jargon-heavy | "Matched 212 of 215 transactions." | "Leveraging AI-driven reconciliation synergies." |
Messaging hierarchy
Tagline one public line "Books closed by Friday."
Value proposition who + outcome + how, 1–2 lines for firms of 1–10 bookkeepers
Message pillars 3 themes, each a benefit speed · accuracy · calm
Proof points 2–3 per pillar: numbers, "matches 95% of lines"
features, customers, awards (only if true and sourced)
Boilerplate 60–100 word "About" for press and partnersNaming
| Name type | Example pattern | Trademark strength (US) | Trade-off |
|---|---|---|---|
| Descriptive | says what it does ("Best Plumbing") | weak; hard to register without acquired distinctiveness | instantly clear, impossible to own |
| Suggestive | hints at a benefit ("Netflix") | strong | needs a moment's thought; often taken |
| Arbitrary | a real word unrelated to the product ("Apple" for computers) | strong | meaning must be built by marketing |
| Coined (fanciful) | invented word ("Kodak") | strongest | empty until you fill it; spelling issues |
| Compound / portmanteau | two words joined or blended | varies | many obvious ones are registered |
| Acronym / initials | letters | weak to moderate | forgettable unless already famous |
| Founder / place name | a surname, a town | varies; surnames can be weak | personal; hard to sell the company later |
The USPTO describes this spectrum from fanciful and arbitrary (strongest) through suggestive to descriptive (weak) and generic (not protectable). A model can generate hundreds of names; it cannot tell you which are free to use.
SCREENING CHECKLIST (the model helps with 1–5; 6–8 are your job)
1. Meets the naming brief (meaning, tone, length)
2. Easy to say and spell after hearing it once
3. No unfortunate meanings or slang in target languages
4. Doesn't describe a feature you might drop
5. Not confusingly close to a competitor's name
6. Domain and social handles available (check registrars yourself)
7. Trademark search in your classes and countries
(USPTO search, WIPO Global Brand Database, EUIPO TMview)
8. Professional clearance by a trademark attorney before launchVisual identity
Logo types
| Type | Best when | Watch |
|---|---|---|
| Wordmark | short, distinctive name; new brand that needs name recognition | long names get tiny at small sizes |
| Lettermark | long name, or initials already used | initials alone carry no meaning for a new brand |
| Symbol (pictorial or abstract) | brand will appear without words: app icons, favicons, products | a new symbol means nothing until people learn it |
| Combination mark | most new brands: the word teaches the symbol | needs rules for when each part appears alone |
| Emblem | heritage, education, sports, craft goods | detail fills in at small sizes; often needs a simplified version |
The logo system
A professional delivers a system, not a file:
| Part | What to specify |
|---|---|
| Primary lockup | the default arrangement, usually horizontal |
| Alternate lockups | stacked, symbol-only, wordmark-only; when each is used |
| Color versions | full color, one color, black, white (reversed) on dark |
| Clear space | margin defined by a part of the logo (e.g. the height of one letter) |
| Minimum size | in px for screens and mm for print; a simplified mark below it |
| Misuse | don't stretch, recolor, outline, add effects, rotate or place on busy images |
| Files | vector masters (SVG, PDF/EPS) plus PNG exports; favicon and app-icon sizes |
Color, type, imagery and icons
| Element | Roles and rules | Ask the model for |
|---|---|---|
| Color palette | primary (recognition), secondary (support), accent (calls to action, sparingly), neutrals (most surfaces), semantic (success, warning, error); give HEX/RGB for screen, CMYK and a spot color for print | roles, usage proportions, accessible text pairs (WCAG 2.2 AA: 4.5:1 body text, 3:1 large text and UI) |
| Typography | a display face for headlines, a text face for body, sometimes a mono or a superfamily that does both; check the license covers web, app and print | 3 pairings with reasons; fallback system fonts; a type scale |
| Imagery | subject, lighting, framing, color treatment, diversity, what never appears | written art direction you can paste into an image tool |
| Iconography | grid size, stroke weight, corner radius, filled vs. outline, metaphor rules | an icon spec and a list of the first 20 icons needed |
See color theory for contrast checks and tonal scales.
Image tools for logo exploration
Image generators are good for fast, cheap exploration: dozens of directions in an afternoon, to react to and to brief a designer with. They are poor at the final mark.
| They do well | They do badly |
|---|---|
| mood boards and visual directions | spelling: lettering is often garbled or inconsistent |
| rough symbol ideas and silhouettes | exact repeats: the "same" logo drifts between images |
| testing a mark in context (sign, cup, app icon mockups) | clean geometry, consistent stroke weights, optical balance |
| style exploration (flat, line, geometric, hand-drawn) | vector output: most produce pixels, not editable shapes |
Prompting them. Describe the mark, not the brand's feelings: type of logo, shape language, style, color count, background, and what to exclude. Generate in batches and vary one thing at a time.
flat vector-style logo symbol, a single abstract shape suggesting a
folded page and a checkmark, geometric, thick even strokes, one color
(dark navy) on a plain white background, centered, no text, no
gradients, no shadows, no 3D, no mockup, simple enough to read at
favicon sizeWhy the final logo must be vector and redrawn by a designer:
- A logo is used from a 16 px favicon to a building sign. Raster images blur or pixelate when scaled; vectors don't.
- Auto-tracing a raster image produces lumpy curves with hundreds of points. A designer rebuilds the mark with clean geometry, consistent stroke weights and optical corrections (circles slightly larger than squares to look equal).
- The wordmark should be set in real, licensed type (or custom-drawn letters), never generated lettering.
- Print needs exact color specs and single-color versions, which a generator won't give you.
- Some tools now offer vector output (e.g. Adobe Illustrator's text-to-vector feature), but the paths still need cleanup, and features change quickly.
Guidelines, architecture and rebrands
What brand guidelines contain
guidelines/01-strategy/ # positioning, audience, promise02-personality/ # traits, archetype, voice, tone03-messaging/ # tagline, pillars, proof04-logo/ # lockups, clear space, misuse05-color/ # roles, values, contrast pairs06-typography/ # faces, scale, licenses07-imagery/ # photo and illustration rules08-iconography/ # grid, stroke, examples09-applications/ # site, slides, email, packaging10-assets/ # files and where to get themSmall teams need a short version: one page each for logo, color, type, voice, plus ten real examples. Long guidelines nobody opens are worse than a one-pager everyone uses.
Brand architecture
David Aaker and Erich Joachimsthaler's brand relationship spectrum (2000) runs from one master brand to many independent ones.
| Model | How it works | Example pattern | Choose it when |
|---|---|---|---|
| Branded house | one master brand; products are descriptors | "FedEx Express", "FedEx Ground" | products share audience and promise; you want every launch to build one name |
| Sub-brands | master brand plus a named product with its own character | "Master + ProductName" | a product needs distinct personality but benefits from the parent |
| Endorsed brands | an independent brand, vouched for by the parent | "Courtyard by Marriott" | the parent adds credibility but would constrain the product |
| House of brands | separate brands, parent mostly invisible to buyers | Procter & Gamble's Tide, Pampers | products serve conflicting audiences or price points; risk must be isolated |
Rebrand or refresh
| Refresh | Rebrand | |
|---|---|---|
| What changes | logo polish, palette, type, photography, templates | strategy, often name and identity |
| Keeps | recognition, core assets | little; equity has to be rebuilt |
| Good reasons | dated look, inconsistent use, new channels | merger, new market or audience, name conflict, reputation problem, strategy has moved |
| Bad reasons | a new executive's taste, boredom | the team is tired of it (customers usually aren't) |
| Risk | low | high: lost recognition, confused customers, cost across every touchpoint |
Brief the model
Each input has a slot in the prompt skeleton from fundamentals; the Slot column says where it goes.
| Input | Why it matters | Example | Slot |
|---|---|---|---|
| What they all say | lets the model find white space | "Everyone says 'simple', 'powerful', 'AI'." (paste their headlines) | <materials> |
| What you do, in one plain sentence | the model fills gaps with generic startup language | "We sell bookkeeping software to firms with 1–10 bookkeepers." | <context> |
| Audience, specific | personality and voice depend on who listens | "Owners of small accounting firms, 40–60, skeptical of hype." | <context> |
| Category and competitors | difference is relative; without names you get clichés | "Competitors: Xero, QuickBooks, spreadsheets, a part-time assistant." | <context> |
| Desired perception | the gut feeling you're aiming at | "After a month, customers should feel 'I'm in control'." | <context> |
| Proof you actually have | stops invented claims | "Founded by two former bookkeepers; 300 paying firms; 4.8 app-store rating." | <context> |
| Hard usage contexts | drives logo type and detail level | "App icon, invoice footer, embroidered polo shirts, one-color fax." | <context> |
| Existing assets and equity | so it doesn't throw away what works | "Customers know our green; keep it." | <context> |
| Markets and languages | names and colors carry meanings | "UK, US and Germany; must work in German." | <context> |
| References and why | taste is easier to show than describe | "We like Mailchimp's voice for its warmth, not its jokes." | <examples> |
| Personality with "but not" | a trait without a limit drifts | "Confident but not arrogant; plain-spoken but not cold." | <requirements> |
| Constraints | budget, timeline, tools, who executes | "No designer yet; I'll use Figma and an image generator." | <requirements> |
| Decision criteria and decider | the model can rank options against them | "The two founders decide; criteria: memorable, ownable, calm." | <quality_bar> |
| Output format | saves reformatting | "A table, then a one-page summary." | <output_format> |
What to ask for
| Stage | Deliverables to ask for by name |
|---|---|
| Discovery | brand audit (how the brand shows up today, touchpoint by touchpoint); competitor audit (claims, colors, archetypes); stakeholder interview guide; customer interview guide; summary of themes |
| Strategy | positioning statement; Dunford positioning worksheet; brand platform (purpose, promise, values, personality); competitive map; archetype analysis; messaging hierarchy |
| Verbal identity | naming brief; name longlist grouped by type; shortlist with rationale and risks; voice chart with "this, not that"; tone by situation; tagline options; boilerplate |
| Visual identity | design brief for a designer; three mood board directions in words; logo exploration prompts; palette with roles and contrast pairs; type pairings; art direction; icon spec |
| Rollout | guidelines outline; touchpoint inventory with priorities; launch announcement; internal FAQ; customer FAQ for a rename |
| Review | critique against the strategy; misuse examples; a test plan with target customers |
The prompt chain
The main deliverable is a brand platform for a product: positioning, personality and voice, a messaging hierarchy, then a direction for the visual identity that a designer can work from. Run stages 1–3 in one chat. Run stage 4 in a fresh chat, because a model reviewing its own draft shares its blind spots.
1. Interview
Use first, when you have notes about the product and customers but no agreed strategy.
<context>
I need a brand platform for [business, e.g. bookkeeping software for small accounting firms]: positioning, personality and voice, a messaging hierarchy, then a direction for the visual identity.
Stage: [stage: new brand | refresh | rebrand]. I am not a brand strategist.
</context>
<task>
Interview me first: ask the questions a brand strategist asks in discovery, before any strategy is written.
</task>
<requirements>
- At most 10 questions, grouped under: the business and audience, competitive alternatives (including doing nothing), proof, desired perception and personality, hard usage contexts, constraints and who decides.
- For each question: why it matters to the platform, and 2-3 typical answers I can pick from or adapt.
- Ask explicitly what proof I have (numbers, customers, history); I may not know what counts.
- Do not draft positioning, names or visuals yet.
</requirements>
<process>
Ask the questions, then stop and wait for my answers.
</process>2. Options
Use once the interview is answered, to choose a positioning territory before anything is built on it.
<materials>
[paste the interview questions and my answers, and competitors' homepage headlines if you have them]
</materials>
<task>
Propose 3 materially different positioning territories for this brand.
</task>
<requirements>
- Work in April Dunford's order: competitive alternatives first, then unique attributes, value and proof, best-fit audience, market category.
- For each territory: the alternative it positions against, the category frame and what it implies about price, the point of difference and the proof behind it, a fitting archetype (as a briefing shorthand, not research), which archetypes competitors already use, and the main risk.
- Mark any point of difference with no proof in my answers [NEEDS PROOF].
- Recommend one and say what would have to be true for each of the others to win.
</requirements>
<process>
Give the territories as a table, then the recommendation, then stop and wait for my choice.
</process>3. Produce
Use after you have chosen a territory: the full skeleton, filled in for the platform.
<materials>
Interview answers and the territory I chose: [paste]
Competitors' claims: [paste their homepage headlines, taglines and About paragraphs]
</materials>
<context>
Business: [one plain sentence, e.g. "We sell bookkeeping software to firms with 1-10 bookkeepers."]
Audience: [audience, e.g. owners of small accounting firms, 40-60, skeptical of hype]
Desired feeling after a month: [feeling, e.g. "I'm in control"]
Proof we have: [real facts only, e.g. founded by two former bookkeepers; 300 paying firms]
Hard uses: [contexts, e.g. app icon, invoice footer, embroidered shirt, one-color stamp]
Stage: [stage: new brand | refresh | rebrand]
Logo type to explore first: [logo type: wordmark | lettermark | symbol | combination mark | emblem | let the strategy decide]
The platform is for [readers, e.g. the two founders, who decide, and a freelance designer who will be briefed from it].
</context>
<examples>
[optional: 1-3 brands outside our category whose voice or identity you admire, and exactly what you admire, e.g. "Mailchimp's voice: the warmth, not the jokes"; 1 you don't want to resemble, and why]
</examples>
<task>
A brand platform: positioning, promise, personality and voice, messaging hierarchy, then a direction for the visual identity that a designer can work from.
</task>
<requirements>
- Positioning: a statement in Geoffrey Moore's format, built from Dunford's five components, against the alternative I chose, plus a one-line homepage version.
- Brand promise: one sentence a support agent could keep.
- Personality: 3-5 traits, each with a "but not"; one archetype from Mark and Pearson's twelve as a shorthand, presented as a framework, not as research.
- Voice: a position on Nielsen Norman Group's four tone dimensions (funny-serious, formal-casual, respectful-irreverent, enthusiastic-matter-of-fact), a "we are / we are not" table with a sample line for each, and the tone for an error, a price increase and a celebration.
- Messaging hierarchy: 5 tagline options, a value proposition, 3 pillars, 2-3 proof points per pillar, and a 60-100 word boilerplate.
- Identity direction: logo type and why, shape language, palette roles (primary, secondary, accent, neutrals), type direction and imagery rules, each traced to a line of the strategy. No finished logo.
- Never invent numbers, names, quotes or sources: write [NEEDS PROOF] where a claim has no proof I supplied.
- Do not say any name or mark is available or trademark-free.
- End with a list of the assumptions you made.
</requirements>
<quality_bar>
Before answering, check your draft against:
- The competitor test: put a competitor's name in place of ours; rewrite every line that still works until it couldn't.
- Every pillar and proof point traces to my materials, or is marked [NEEDS PROOF].
- Each "but not" rules out something a competitor does.
- The identity direction survives the hard uses: 16 px, one color, embroidery.
</quality_bar>
<output_format>
Sections in this order: positioning, promise, personality, voice (tables), messaging hierarchy (a table: level | content | proof | source), identity direction (one short paragraph per element). Under 1,500 words, then the assumptions.
</output_format>
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
Write the positioning first and check it against the quality bar before building the rest on it.
</process>4. Review
Use when the platform is back: open a fresh chat, paste the brief and the platform, and ask for a ranked critique, not a rewrite.
<materials>
The brief: [paste the stage 3 prompt]
The brand platform: [paste]
Competitors' homepage headlines: [paste]
</materials>
<context>
This is a fresh chat. The platform below was written in another conversation; review it as someone seeing it for the first time, before it is shared with a designer.
</context>
<task>
A review of this brand platform. Do not rewrite it.
</task>
<requirements>
- Judge against: strategic fit, distinctiveness, simplicity, scalability of the identity direction, ownability, consistency of the voice across situations, honesty of claims, and whether a non-designer could use it.
- Run the competitor test on every line of the positioning and messaging, and list the lines that fail.
- List every claim that needs proof, and say whether the brief supplied it.
- Quote the line behind every finding.
</requirements>
<quality_bar>
Before answering, check that every finding names a criterion, quotes the line, and proposes a specific fix.
</quality_bar>
<output_format>
A table: rank | finding | criterion | quoted line | fix, most important first. Then one line: ready, fix first, or rethink the positioning.
</output_format>
<process>
Rank the issues by how much they weaken the brand; do not produce a revised platform.
</process>Goals and loops
In an agent that runs commands and checks its own work, such as Claude Code, the Produce stage can run until it
measurably passes: /goal keeps starting turns until a condition holds, and /loop repeats a prompt or skill on
an interval. The checker reads only the conversation, so every condition below names a check whose output Claude
prints, the constraints that must survive, and a bound. How to write your own:
Goals and loops.
| Task | Completion condition |
|---|---|
| Naming longlist | /goal names.md lists 20 candidates, each with its naming type and the whois result for its .com printed as available or taken; or stop after 8 turns. Trademark clearance is still a job for you and counsel |
| Brand guidelines | /goal brand-guidelines.md has every section of the template, every color lists hex, RGB and CMYK with its contrast ratio against white and black printed by python contrast.py, and every lockup links an SVG; or stop after 10 turns |
| Voice across pages | /goal vale site/ with a Vale style built from the voice guide reports 0 errors (show its summary), and no page was rewritten beyond the flagged sentences; or stop after 10 turns |
Loops that fit: a monthly /schedule that collects competitors' new messaging, campaigns and rebrands from their sites and summarizes what changed against your positioning map.
What a checker can't judge: whether the brand means anything to the people you sell to. Test names and positioning with real customers.
Prompt library
One entry per deliverable a brand designer or strategist produces, besides the platform in the chain. Quick is for when you're in a hurry; Full is for when the result matters.
Competitor audit
Stage: produce · Use when: you need a point of difference and suspect you sound like everyone else.
Quick
Here are the homepage headlines, taglines and About paragraphs of [number, e.g. 6] competitors: [paste]. Build a table: brand, main claim (quoted), archetype, personality traits, dominant color, typical imagery.
Then list the claims everyone makes, the claims nobody makes, and 3 open positioning territories with the risk of each. Quote; don't paraphrase.Full
<materials>
[paste each competitor's homepage headline, tagline and About paragraph, labeled by brand, plus a short description of each logo and its colors]
</materials>
<context>
We are: [one plain sentence, e.g. "bookkeeping software for firms with 1-10 bookkeepers"]
Our audience: [audience, e.g. owners of small accounting firms]
Alternatives our customers consider: [alternatives, e.g. QuickBooks, spreadsheets, a part-time assistant, doing nothing]
</context>
<task>
A competitor audit that shows where every brand in the category sits and which positioning territories are open to us.
</task>
<requirements>
- For each brand: main claim (quoted), archetype as a shorthand, 3 personality traits, dominant color, typical imagery, logo type.
- The claims everyone makes (points of parity) and the claims nobody makes.
- A competitive map: propose 2 axes buyers actually choose on, and place every alternative on it, including doing nothing.
- 3 open territories, each with the proof we would need to own it and the risk (an empty corner can mean nobody wants it).
- Use only what is in the materials and mark anything you infer as inferred. Never invent numbers, names, quotes or sources: write [NEEDS SOURCE] where a fact is missing.
- End with a list of the assumptions you made.
</requirements>
<quality_bar>
Before answering, check your draft against:
- Every claim attributed to a brand is quoted from the materials.
- The map's axes are things buyers choose on, not things we happen to be good at.
- Each open territory says how we would test that customers want it.
</quality_bar>
<output_format>
A table with one row per brand, the map as a list of coordinates on the two axes, the territories in under 100 words each, then the assumptions.
</output_format>
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>Naming longlist
Stage: produce · Use when: you have a naming brief and want breadth, then a defensible shortlist to take into legal checks.
Quick
Generate 40 names from this naming brief: [paste], 8 each that are suggestive, arbitrary, coined, compound and metaphor-based. For each: one-line rationale, pronunciation and main risk.
Don't claim any name is available; I'll check domains and trademarks myself.Full
<materials>
[optional: paste the naming brief, if you have one]
</materials>
<context>
What it names: [thing: company | product | feature | sub-brand]
Positioning and personality: [paste, e.g. "fast service, calm feeling"; confident but not loud]
Must convey: [ideas, e.g. calm, speed, a pause in the day]
Must avoid: [words, e.g. the ones competitors use: "brew", "bean", "daily"]
Length: [limit, e.g. 2 syllables or under 8 letters]
Markets and languages: [markets, e.g. US, UK and Germany]
Competitors: [names, e.g. the 3 brands customers compare us with]
</context>
<examples>
[optional: 2-3 names you like from any category and why, e.g. "a short real word used in an unexpected way"; 1 you dislike, and why]
</examples>
<task>
A longlist of names that meet the brief, then a scored shortlist of the 10 strongest, to take into domain and trademark checks.
</task>
<requirements>
- 40 names: 8 each that are suggestive, arbitrary, coined, compound and metaphor-based. Label each with its type on the USPTO strength spectrum (fanciful and arbitrary strongest, suggestive next, descriptive weak).
- For each: a one-line rationale, pronunciation, and any risk (a meaning or slang in a target language, likely to be common, too close to a competitor).
- Score the 10 strongest 1-5 on: fits the brief, memorable, easy to say and spell after hearing it once, distinct from competitors, works in every target language, room to grow.
- Do not claim any name is available or trademark-free: write [CHECK REGISTER] beside every shortlisted name.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE] where support is missing.
- End with a list of the assumptions you made.
</requirements>
<quality_bar>
Before answering, check your draft against:
- No name describes a feature we might drop.
- No two shortlisted names are variations of one idea.
- Doubtful meanings in a target language are flagged, not kept silently.
</quality_bar>
<output_format>
The longlist as a table (name | type | rationale | pronunciation | risk), then the shortlist scoring table, then the checks I must run myself (domains, handles, trademark search in my classes and countries, attorney clearance), then the assumptions.
</output_format>
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>Voice and tone guide
Stage: produce · Use when: positioning is set, and copy from different people sounds like different companies.
Quick
Create a voice guide for [brand, e.g. a bookkeeping app] whose personality is [traits, e.g. direct but not curt, warm but not gushing]: a position on NN/g's four tone dimensions, a "we are / we are not" table with sample lines, the tone for an error, a price increase and a celebration, and 10 words we use and 10 we avoid.
Check that every sample line sounds like the same person.Full
<materials>
Positioning and personality: [paste]
Copy to rewrite: [paste a real page or email]
</materials>
<context>
Audience: [audience, e.g. owners of small accounting firms, skeptical of hype]
Who will use the guide: [writers, e.g. two founders, a support team of four, a freelance copywriter]
Channels: [channels, e.g. app interface, support emails, website, social posts]
</context>
<examples>
[paste 3 pieces of copy that sound like us, or like who we want to be, and why; 3 that don't, and why]
</examples>
<task>
A voice and tone guide that people with no writing training can follow.
</task>
<requirements>
- A position on Nielsen Norman Group's four tone dimensions (funny-serious, formal-casual, respectful-irreverent, enthusiastic-matter-of-fact), each 1-5 with a reason.
- A "we are / we are not" table, 3-5 rows, each with a line that sounds like us and one that doesn't.
- Tone by situation: onboarding, error, price increase, outage, celebration, legal notice; one example line each.
- 8 concrete writing rules inferred from the examples (sentence length, person, contractions, jargon, humor, punctuation), plus 10 words we use and 10 we avoid.
- Rewrite the pasted copy with the rules and say which rule drove each change.
- Never invent numbers, names, quotes or sources: write [NEEDS PROOF] beside any claim in the rewrite that I haven't supported.
- End with a list of the assumptions you made.
</requirements>
<quality_bar>
Before answering, check your draft against:
- Every rule can be tested on a sentence: a writer could say whether it was followed.
- The outage and legal lines still sound like the same brand, only more serious.
- The rewrite changes no facts and adds no claims.
</quality_bar>
<output_format>
The dimensions table, the voice table, the tone table, the rules, the word lists, then the rewrite with change notes. Under 1,200 words.
</output_format>
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>Logo exploration prompts
Stage: produce · Use when: the strategy is set and you want visual directions to react to, or to brief a designer with.
Quick
From this strategy: [paste], give 4 logo directions of different kinds (e.g. abstract symbol, pictorial, lettermark, emblem). For each: why it fits, and an image-generator prompt that describes only the mark (shape, stroke, color count, plain background) and ends "no text, no gradients, no shadows, no mockups, no 3D".
Say how each would fail at 16 px.Full
<materials>
Strategy: [paste the positioning, personality and audience]
</materials>
<context>
Name: [name, e.g. Slow Lane]
Hard uses: [contexts, e.g. favicon, a hanging sign seen from 20 m, an embroidered polo shirt, a one-color stamp]
Category clichés to avoid: [symbols, e.g. cups, beans, steam]
These are explorations to react to and to brief a designer with, not the final logo.
</context>
<examples>
[optional: 1-3 marks from outside our category that you admire, described in words (shape, weight, color count) with why; 1 style you don't want. Describe them; don't upload competitors' logos]
</examples>
<task>
Four logo directions, each turned into prompts for an image generator, so I can explore before briefing a designer.
</task>
<requirements>
- The directions differ in kind, e.g. geometric abstract, pictorial, lettermark, emblem.
- For each: the idea in one sentence and the line of the strategy it expresses.
- An image prompt that describes only the mark: logo type, shape language, stroke, color count, plain background, and an exclusion list (no text, no gradients, no shadows, no mockups, no 3D). No lettering in any prompt: image tools garble it, and the wordmark will be set in real type. Describe a lettermark in words for a designer instead.
- 3 variations per direction, each changing one thing.
- How each would fail at 16 px, in one color and embroidered.
- No reference to existing brands or named living designers.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE] where support is missing.
- End with a list of the assumptions you made.
</requirements>
<quality_bar>
Before answering, check your draft against:
- Each mark can be described in one sentence and drawn from memory.
- At least one direction avoids the category's literal symbols.
- No prompt asks for text or a mockup.
</quality_bar>
<output_format>
Per direction: idea, strategy link, prompt (plain text I can paste), 3 variations, failure cases.
</output_format>
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>Designer brief
Stage: produce · Use when: you're hiring a designer and want to brief them like a professional would.
Quick
Write a one-page identity design brief for a freelance designer from this strategy: [paste]. Include the audience, positioning, personality, competitors' visual codes to avoid, hard uses, deliverables (logo system, palette with roles, type pairing, icon style, short guidelines, vector masters), timeline, budget and how options will be judged.
Check that a designer could quote from it without a call.Full
<materials>
Strategy: [paste the brand platform]
Competitors' visual codes: [paste the competitor audit, or describe their colors, marks and imagery]
</materials>
<context>
Stage: [stage: new brand | refresh | rebrand]
Assets to keep: [assets, e.g. "customers know our green", or "none"]
Designer: [designer, e.g. a freelancer found through a portfolio site, briefed by email]
Timeline: [dates, e.g. first concepts in 3 weeks, final files in 8]
Budget range: [range, e.g. $5,000-8,000]
Who decides: [decider, e.g. the two founders]
</context>
<examples>
[1-3 identities from outside our category that you admire, and exactly what, e.g. "the restraint: one color, one typeface"; 1 you don't want to resemble, and why]
</examples>
<task>
A one-page identity design brief, so a designer can quote, plan and start without a discovery call.
</task>
<requirements>
- Sections: background, audience, positioning, personality, competitors and their visual codes to avoid, hard usage contexts, deliverables, what I will provide, timeline, budget, how options will be judged, and who decides.
- Deliverables by name: a logo system (primary and alternate lockups, color versions, clear space, minimum size), a palette with roles, a type pairing with licenses covering web, app and print, an icon style, short guidelines, and vector masters (SVG, PDF) plus PNG exports.
- Judging criteria drawn from the strategy, not taste, e.g. distinct from competitors, legible at 16 px, works in one color.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE] where support is missing.
- End with a list of the assumptions you made.
</requirements>
<quality_bar>
Before answering, check your draft against:
- It fits on one page, about 500 words.
- Someone other than me could check every judging criterion on a concept.
- It asks for a system with rules, not one mark.
</quality_bar>
<output_format>
One page with the sections above as short headings, then the assumptions.
</output_format>
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>Brand guidelines
Stage: produce · Use when: an identity exists and a small team needs rules they will actually read.
Quick
Write short brand guidelines for a [team, e.g. 5-person team with no designer] from these inputs: [paste]. Sections: strategy (one page), logo (lockups, clear space, minimum size, 6 misuse examples), color, type, imagery, voice, 10 application examples.
Keep each section under 200 words, each with a "when in doubt" rule.Full
<materials>
[paste the positioning, personality, voice guide, a description of the logo files, palette HEX values and typefaces]
</materials>
<context>
Team: [team, e.g. 5 people, no in-house designer, making slides, emails and social posts]
Brand architecture: [architecture: single brand | branded house | sub-brands | endorsed brands | house of brands]
What the team makes most: [list, e.g. sales slides, invoices, social posts]
</context>
<task>
Short brand guidelines a non-designer can make a correct slide from.
</task>
<requirements>
- Sections: strategy on one page; logo (lockups, clear space, minimum size in px and mm, color versions, 6 misuse examples); color (roles, values, usage share, accessible text pairs); typography (faces, weights, scale, fallbacks, license notes); imagery; voice; 10 application examples (email signature, slide, social post, invoice and so on).
- Text color pairs must meet WCAG 2.2 AA (4.5:1 for body text, 3:1 for large text and UI); write [VERIFY] beside every ratio so I check it with a contrast tool.
- Each section under 200 words, with one "when in doubt" rule.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE] where support is missing.
- End with a list of the assumptions you made.
</requirements>
<quality_bar>
Before answering, check your draft against:
- Every rule comes with an example of correct use, and the logo section with examples of misuse.
- Nothing contradicts the strategy or the voice guide.
- The whole document can be read in 10 minutes.
</quality_bar>
<output_format>
Markdown with one heading per section and tables for color and type, then the assumptions.
</output_format>
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>Refresh or rebrand
Stage: options · Use when: someone proposes a new name or look and you need a reasoned decision before any design work.
Quick
We're considering [change: refresh | rebrand | new name] because [reason, e.g. a merger, a name conflict, a dated look]. Argue for a refresh, a full rebrand and no change: equity kept or lost, cost across our touchpoints, risks.
Recommend one, say what would have to be true for the others, and stop there.Full
<materials>
[paste what customers say about us, any recognition data, and the list of touchpoints: site, app, invoices, signage, packaging, vehicles]
</materials>
<context>
Why change is proposed: [reason: dated look | inconsistent use | merger | new market or audience | name conflict | reputation problem | new leadership's taste]
What customers know us for: [assets, e.g. the green, the owl symbol, the name]
Who decides, and when: [decider, e.g. the board, in 6 weeks]
</context>
<task>
A reasoned recommendation on whether to refresh, rebrand or change nothing, so the decision rests on strategy rather than taste.
</task>
<requirements>
- Argue each option in turn, at its strongest: equity kept or lost, cost and effort across the listed touchpoints, risks, and timeline.
- Separate good reasons for change (the strategy has moved, a name conflict, a merger) from bad ones (boredom, a new executive's taste), and say which this case is.
- If the name may change, list the legal checks needed and who does them.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE] where support is missing.
- End with a list of the assumptions you made.
</requirements>
<quality_bar>
Before answering, check your draft against:
- Each option gets its strongest case, not a straw man.
- The recommendation names the evidence that would change it.
- Costs are listed per touchpoint, not as "significant".
</quality_bar>
<output_format>
A comparison table (option | equity kept | equity lost | cost by touchpoint | risk), the recommendation in under 150 words, then the assumptions.
</output_format>
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on. Give the options and the recommendation, then stop; don't plan the rollout yet.
</process>Logo concept critique
Stage: review · Use when: you have a sketch or generated image and want to know what a professional would change.
Quick
[attach image] An early logo concept for [brand, e.g. a coffee shop by a commuter station], used on [contexts, e.g. cup sleeve, hanging sign, app icon]. Critique it: concept clarity, what it resembles, legibility at 16 px and at 20 m, one-color and reversed versions, stroke consistency, and what must change when it is redrawn as vector.
Rank the fixes; don't redraw it.Full
<materials>
[attach the image or images]
Strategy: [paste the positioning and personality]
</materials>
<context>
Brand: [brand, e.g. Slow Lane, a coffee shop by a commuter station]
Hard uses: [contexts, e.g. cup sleeve, a hanging sign seen from 20 m, app icon, one-color rubber stamp]
Where it came from: [source: image generator | sketch | a designer's first round]
</context>
<task>
A critique of this logo concept before it goes further, so I know what a designer would change. Do not redraw it.
</task>
<requirements>
- Judge against the rubric: strategic fit, distinctiveness, simplicity, scalability, ownability.
- Check: the concept in one sentence; common category symbols or existing logos it resembles; legibility at 16 px and at 20 m; one-color, reversed and embroidered versions; balance and stroke consistency; any generated lettering.
- Say what must change when it is redrawn as vector with licensed type.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE] where support is missing. Name a resembling logo only if you are confident, and recommend a reverse image search either way.
- End with a list of the assumptions you made.
</requirements>
<quality_bar>
Before answering, check your draft against:
- Every finding points at a specific part of the image.
- The fixes are ranked by how much they change the result, not by how easy they are.
- No finding is taste alone, without a criterion behind it.
</quality_bar>
<output_format>
A table: rank | finding | criterion | fix. Then 3 revision directions, one sentence each.
</output_format>
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>Worked example
Weak prompt
Make me a logo for my coffee shop. Something modern and cool.What's wrong
- No strategy: who it's for and how it differs, so every direction is a guess.
- "Modern and cool" means nothing to a designer or a model; it produces the category average.
- No hard uses (sign, cup, app icon, stamp), so nothing forces a simple mark.
- It asks an image tool for a finished logo, which it can't deliver: the lettering comes out garbled and the file is pixels.
Strong prompt
<context>
I'm opening a coffee shop called "Slow Lane" next to a commuter train station.
Audience: commuters 25-45 who want a calm 3 minutes, not a lounge.
Competitors: a chain cafe in the station (fast, impersonal) and an artisan cafe 1 km away (slow, expensive).
We want to own "fast service, calm feeling". Proof we have: none yet; we open in March.
Hard uses: a cup sleeve, a hanging sign seen from 20 m, a loyalty app icon, a rubber stamp in one color.
</context>
<examples>
I like the restraint of old railway signage: one color, bold shapes, no decoration. I don't want the cozy hand-drawn chalkboard look every cafe nearby has.
</examples>
<task>
A short brand strategy and three logo directions in words, each with an image-generator prompt, so I can explore before hiring a designer.
</task>
<requirements>
- A positioning statement in Geoffrey Moore's format and 4 personality traits, each with a "but not".
- Three logo directions of different kinds (e.g. symbol, lettermark, emblem), each traced to the positioning.
- Each image prompt describes only the mark: shape, stroke, color count, plain background; it excludes text, gradients, shadows and mockups.
- Never invent numbers, names, quotes or sources: write [NEEDS PROOF] for any claim we can't back yet.
- End with a list of the assumptions you made.
</requirements>
<quality_bar>
Before answering, check your draft against:
- The competitor test: neither cafe could use these lines or marks.
- Every mark survives favicon size and one color.
- No image prompt asks for lettering.
</quality_bar>
<output_format>
The positioning, the traits as a table, then per direction: idea, why it fits, image prompt. Under 500 words.
</output_format>
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>What came back
Illustrative, abridged:
Positioning: For commuters who want a calm moment on the way to the train, Slow Lane is a station coffee bar that serves you in under 3 minutes [NEEDS PROOF] in a quiet room. Unlike the chain cafe, it feels personal; unlike the artisan cafe, it won't make you miss your train.
Traits: calm, but not sleepy · quick, but not rushed · warm, but not cutesy · plain-spoken, but not blunt
Direction 1, symbol: a cup whose steam is a winding road.
Prompt: flat vector-style logo symbol, a single continuous line forming a steaming cup that doubles as a winding road, thick even stroke, rounded ends, one color (deep green) on plain white, centered, no text, no gradients, no shadows, no 3D, no mockup
Direction 2, lettermark: "SL" as two lanes merging.
Prompt: minimal monogram logo, the letters "SL" in a bold geometric sans serif, the S flowing into the L like two road lanes merging, one color (deep green), plain white background, no gradients, no mockup
Direction 3, emblem: the name around a ring like a station clock face …
Assumptions: deep green reads as calm; the sign is lit at night.How an expert judged it
- Strategic fit: pass. The positioning names both alternatives and the tension; "won't make you miss your train" is a line neither competitor could say.
- Personality: pass. Each "but not" rules out something a competitor does.
- Honesty: pass. The model marked the under-3-minutes claim
[NEEDS PROOF]instead of asserting it; time real orders before using it. - Image prompts: fail on direction 2. The prompt asks the image tool to draw the letters "SL". Generators garble or vary letterforms, and a lettermark's whole value is in drawn letters, so this direction should be described in words for a designer, not generated. It also breaks the third quality-bar check: read the prompts against your bar, not just the pictures they make.
- Scalability: weak on direction 3. An emblem with the name around a ring won't survive the app icon or the stamp; it needs a simplified version.
The follow-up
Direction 2's prompt asks the image tool to draw the letters "SL", which breaks the no-lettering check. Keep directions 1 and 3 as they are. Replace direction 2's prompt with a written brief a designer could use to draw the lettermark: the idea, how the letters are built, stroke weight relative to the symbol in direction 1, and how it must read at 16 px. For direction 3, add a simplified version for the app icon and the stamp. Change nothing else.Judge the output
| Criterion | Good | Red flag |
|---|---|---|
| Strategic fit | each choice traces back to audience, difference or personality | "modern, clean, innovative" with no link to the strategy |
| Distinctiveness | wouldn't fit a named competitor if you swapped the name | the competitor test passes: it could be anyone's |
| Simplicity | the mark can be described in one sentence and drawn from memory | multiple ideas crammed into one symbol |
| Scalability | legible at 16 px and in one color, works reversed | thin lines, tiny details, gradients doing the work |
| Ownability | name and mark are unusual in the category and pass your searches | a dictionary word in your own category; a common symbol (leaf, swoosh, globe) |
| Consistency | voice examples sound like the same person in every situation | tone swings or turns into a different brand under stress |
| Honesty | every claim has a reason to believe you supplied | invented stats, awards or customer counts |
| Usability of guidelines | a non-designer can make a correct slide from them | rules without examples, or 80 pages |
The same checks are built into every <quality_bar> on this sheet, so the model applies them before you see a draft;
use the table to check what comes back. The quickest test of distinctiveness:
Put [competitor, e.g. the station chain cafe]'s name in place of ours in this platform. Which lines still work? List them with the reason; don't rewrite yet.Pitfalls
| Pitfall | Why it hurts | Fix |
|---|---|---|
| Starting with the logo | visuals with no strategy are taste contests | strategy first; brief visuals from it |
| Adjective soup ("innovative, trusted, modern") | every brand says it; the model echoes it | traits with "but not", plus examples of sound-like and not |
| Believing "this name appears available" | the model has no live trademark or domain data | search registers yourself; hire an attorney for clearance |
| Using a generated image as the final logo | pixels, garbled lettering, look-alike risk, weak copyright position | use it to explore; have a designer redraw in vector with licensed type |
| Literal symbols (a cup for coffee, a house for real estate) | everyone in the category uses them | ask for abstract or metaphorical directions too |
| Designing only for the ideal case | the logo breaks on a favicon, invoice or embroidery | list hard uses in every prompt |
| Too many colors and typefaces | inconsistent, expensive, hard to use | one or two typefaces; palette with roles and proportions |
| Copying a famous brand's style | looks derivative; legal risk | reference qualities ("warm, plain-spoken"), not brands to copy |
| Rebranding out of boredom | throws away recognition customers rely on | refresh first; rebrand only for a strategic reason |
| Guidelines nobody reads | the system decays within months | short, example-led, with templates |
| Letting the model invent proof | false claims damage trust and can break advertising law | supply proof; ask it to flag claims needing evidence |
References
- Marty Neumeier, The Brand Gap (opens in a new tab) (New Riders, 2003): the "gut feeling" definition; differentiation, collaboration, innovation, validation, cultivation
- Alina Wheeler with Rob Meyerson, Designing Brand Identity, 6th ed. (opens in a new tab) (Wiley, 2024): the standard reference on identity process, logo types and guidelines
- April Dunford, Obviously Awesome (opens in a new tab) (2019): the five components of positioning and the order to work them
- Nielsen Norman Group, "The Four Dimensions of Tone of Voice" (opens in a new tab): funny–serious, formal–casual, respectful–irreverent, enthusiastic–matter-of-fact
- Nielsen Norman Group, "The Impact of Tone of Voice on Users' Brand Perception" (opens in a new tab): a study of how different tones change users' perceptions of the same brand
- Mailchimp Content Style Guide, Voice and Tone (opens in a new tab): a public example of a voice that stays constant and a tone that adapts
- USPTO, Trademark basics (opens in a new tab): what trademarks are and how registration works
- USPTO, Strong trademarks (opens in a new tab): the fanciful-to-generic strength spectrum
- USPTO, Likelihood of confusion (opens in a new tab): why similar marks for related goods conflict
- USPTO trademark search (opens in a new tab): the US register
- WIPO Global Brand Database (opens in a new tab): trademarks across many national and international registers
- EUIPO TMview (opens in a new tab): trademarks registered in EU and partner offices
- US Copyright Office, Copyright and Artificial Intelligence (opens in a new tab): the report series, including Part 2 on the copyrightability of AI-generated output (January 2025)
- Google Fonts Knowledge (opens in a new tab): practical lessons on choosing and pairing type
- Geoffrey A. Moore, Crossing the Chasm (HarperBusiness, 1991; 3rd ed. 2014): the positioning statement template
- Margaret Mark and Carol S. Pearson, The Hero and the Outlaw (McGraw-Hill, 2001): the twelve brand archetypes
- David A. Aaker and Erich Joachimsthaler, "The Brand Relationship Spectrum" (California Management Review, 2000): branded house to house of brands