Brand Voice Guide#

This skill produces a concrete, example-driven brand voice document that any writer — internal, freelance, or AI — can follow to match the brand's sound on the first try, without needing brand-team sign-off for every sentence. It covers voice attributes, tone spectrum, vocabulary rules, do/don't examples across channels, and a self-audit checklist.

Quick Reference#

DecisionStrongAcceptableWeakVoice attributes3-4 defined with spectrum scales, each with do/don't examples3-4 attributes listed with brief descriptionsVague terms like "professional" with no examplesVocabulary list50+ terms with approved/banned alternatives and context notes20-30 approved/banned termsNo vocabulary guidance or just a few random wordsTone spectrum5+ scenarios mapped on a casual-to-formal scale with examplesGeneral guidance on when to be formal vs. casualSingle tone for all contextsChannel adaptationSeparate guidelines per channel (email, social, ads, CX, product) with examplesGeneral channel notes without examplesOne voice doc applied identically everywhereDo/don't examples10+ paired examples per channel showing exact rewrites5-10 generic do/don't pairsAbstract rules with no concrete examplesAudience awarenessVoice adjusted by segment (new vs. loyal, B2B vs. B2C)Single audience definitionNo audience considerationTesting methodBefore/after audit of real brand copy with scoring rubricInformal review by brand teamNo validation processAI prompt integrationVoice attributes formatted as system prompt instructionsGeneral AI usage notesNo AI guidance

Solves#

1. Inconsistent brand voice across writers — five freelancers, three agencies, and two internal teams all writing in different voices because no concrete reference exists beyond "be authentic."
2. Brand voice that only lives in the founder's head — the founder approves every piece of copy because no one else can match the tone, creating a bottleneck that limits content velocity.
3. Tone-deaf channel adaptations — the same formal product-page voice used in Instagram captions and customer support replies, making the brand feel robotic on casual channels.
4. AI-generated content that sounds generic — ChatGPT or Claude outputs that are technically correct but sound like every other brand because the prompts lack specific voice constraints.
5. New hire ramp-up time — copywriters and CX reps taking months to internalize brand voice because training is tribal knowledge, not documented.
6. Brand drift over time — voice evolving randomly as new people join and old ones leave, with no anchor document to course-correct against.
7. Vocabulary inconsistencies — using "clients" in one email and "customers" in the next, or mixing "we" and "the company" within the same page.

Workflow#

Step 1 — Audit Existing Voice#

Collect and analyze the brand's current voice across channels:

• Gather 20-30 samples of existing brand copy: website pages, emails, social posts, ads, CX replies, product descriptions
• Identify what sounds "most like the brand" vs. what sounds off-brand (ask the founder or brand lead to rank them)
• Note recurring patterns: sentence length, punctuation habits, emoji usage, contractions, jargon, humor level
• Document any existing brand guidelines, style guides, or voice notes
• Review competitor voices to identify differentiation opportunities

Deliverable: Voice audit summary with ranked copy samples and pattern analysis.

Step 2 — Define Voice Attributes#

Establish 3-4 core voice attributes that define the brand's personality:

Structure for each attribute:

• Attribute name (e.g., "Confident," "Warm," "Direct," "Playful")
• Spectrum scale showing where the brand sits (e.g., Confident: 7/10 — assertive but never arrogant)
• "This means..." statement explaining the attribute in practice
• "This doesn't mean..." statement clarifying the boundary
• 3 do/don't pairs showing the attribute in action

Attribute selection principles:

• Choose attributes that differentiate — "professional" is too generic
• Each attribute should create tension with at least one other (e.g., "confident" vs. "humble" forces intentional choices)
• Attributes should be testable — you can look at a sentence and judge whether it embodies the attribute
• Avoid aspirational attributes that don't match current brand reality

Deliverable: Voice attribute cards with spectrum scales, definitions, and example pairs.

Step 3 — Build Vocabulary Lists#

Create approved and banned word lists organized by category:

Categories to cover:

• Brand-specific terms (product names, feature names, brand values)
• Customer references ("customers" vs. "clients" vs. "members" vs. "community")
• Self-references ("we" vs. company name vs. "our team" vs. "I")
• Action words (preferred verbs for CTAs, feature descriptions, benefits)
• Banned words (competitor names, outdated terms, jargon to avoid, words that contradict voice attributes)
• Sensitive terms (inclusive language preferences, accessibility-first language)
• Industry jargon (which technical terms to use vs. simplify)

Format for each entry:

• Approved term → context where it's used → example sentence
• Banned term → why it's banned → approved alternative

Deliverable: Vocabulary reference with 50+ entries across all categories.

Step 4 — Map Tone Spectrum#

Define how voice tone shifts across different contexts while maintaining the same underlying personality:

ContextTone ShiftExampleProduct launch announcementPeak excitement, confidentError message / outage notificationCalm, direct, empatheticCustomer complaint responseWarm, accountable, solution-focusedSales emailEnthusiastic but not pushyLegal / complianceClear, formal, still humanSocial media comment replyCasual, quick, personality-forwardOnboarding sequenceEncouraging, helpful, patientWin-back / re-engagementPersonal, low-pressure, value-focused

For each context, provide a before (wrong tone) and after (right tone) rewrite of a realistic example.

Deliverable: Tone spectrum matrix with scenario-specific guidelines and example rewrites.

Step 5 — Write Channel-Specific Guidelines#

Create detailed voice adaptations for each brand channel:

For each channel, document:

• Tone adjustments from the base voice (what dials up, what dials down)
• Formatting conventions (emoji policy, hashtag usage, capitalization, punctuation)
• Length guidelines (sentence length, paragraph length, total length)
• 5+ do/don't example pairs using real or realistic copy
• Common templates with voice guidelines applied (subject lines, CTAs, openings, closings)

Channels to cover:

• Website (homepage, product pages, about page, blog)
• Email (marketing, transactional, CX)
• Social media (platform-specific: Instagram, Twitter/X, LinkedIn, TikTok)
• Advertising (paid social, search ads, display)
• Customer support (live chat, email, knowledge base)
• Product UI (microcopy, tooltips, error messages, success states)

Deliverable: Channel guideline cards with examples, templates, and formatting rules.

Step 6 — Create AI Prompt Instructions#

Translate voice attributes into structured instructions for AI writing tools:

• System prompt block that captures voice attributes, tone, and vocabulary
• Channel-specific prompt modifiers
• Example few-shot pairs showing input → desired output
• Negative examples showing what to reject
• Quality check prompts for self-review

Deliverable: AI prompt template library with system prompts and few-shot examples.

Step 7 — Build Audit and Scoring Rubric#

Create a scoring system to evaluate whether content matches brand voice:

• Score each voice attribute on a 1-5 scale
• Include vocabulary compliance check
• Include tone-appropriateness check for the specific channel/context
• Define minimum passing score
• Create a quick 30-second self-audit checklist for writers

Deliverable: Voice scoring rubric with minimum thresholds and self-audit checklist.

Examples#

Example 1 — DTC Wellness Brand (Warm, Confident, Playful)#

Input: "We're a wellness brand selling supplements and self-care products DTC. Our founder writes all the copy and it sounds great, but we're hiring two freelancers and need to hand off the voice. We want to sound like a knowledgeable friend, not a clinical brand."

Voice attributes defined:

• Warm (8/10): Speak like a friend who happens to know a lot about wellness. Use contractions, ask questions, share openly.
  • Do: "Your skin's telling you something — here's what we'd try first."
  • Don't: "Our dermatologically tested formula addresses multiple skin concerns."
• Confident (7/10): State benefits directly without hedging. Back claims with specifics, not superlatives.
  • Do: "This serum delivers visible results in 14 days — 89% of testers agreed."
  • Don't: "This might be the best serum you've probably ever tried!"
• Playful (6/10): Use light humor and personality, especially in headers and social. Never force it.
  • Do: "SPF that doesn't leave you looking like a ghost. You're welcome."
  • Don't: "LOL this sunscreen is literally THE BEST OMG."

Vocabulary excerpt:

ApprovedBannedContext"ritual""routine"Elevates daily habits"your body""the body"Personal, not clinical"ingredients you recognize""clean ingredients"Specific over buzzwords"community""customers"Relationship over transaction

Tone spectrum applied:

• Product page: Confident + Warm (lead with benefits, back with ingredients)
• Instagram caption: Playful + Warm (personality first, education second)
• CX reply to complaint: Warm + Confident (empathize first, solve fast)
• Email subject line: Playful + Confident (curiosity-driven, specific)

Example 2 — B2B SaaS Platform (Direct, Smart, Human)#

Input: "We sell project management software to mid-market teams. Our voice has drifted — marketing sounds like a startup, sales decks sound corporate, and support sounds robotic. We need one voice that works across all three."

Voice attributes defined:

• Direct (9/10): Get to the point. Lead with the answer, not the context. Respect the reader's time.
  • Do: "Import your Jira board in 3 clicks. Here's how."
  • Don't: "We understand that migrating from one project management tool to another can be a complex and time-consuming process."
• Smart (7/10): Show expertise through specificity, not jargon. Assume the reader is intelligent.
  • Do: "Reduce sprint planning from 2 hours to 20 minutes."
  • Don't: "Leverage our AI-powered agile methodology optimization engine."
• Human (6/10): Sound like a person, not a press release. Use contractions, occasional humor, conversational asides.
  • Do: "Yes, it integrates with Slack. No, it won't spam your channels."
  • Don't: "Our platform offers seamless integration capabilities with leading communication tools."

AI prompt template created:

text

You are writing as [Brand]. Follow these voice rules:
- Be direct: lead with the answer, then explain. No throat-clearing.
- Be specific: use numbers, timeframes, and concrete outcomes.
- Sound human: use contractions, occasional dry humor, and conversational tone.
- Never use: leverage, synergy, best-in-class, cutting-edge, seamless.
- Always use: you/your (not "users"), we (not "the platform").
- Sentence length: average 12-18 words. Max 25.

Common Mistakes#

1. Choosing aspirational over accurate attributes — Picking voice attributes the brand wants to have rather than ones that match its current best copy. Start from what already works and refine, don't invent a new personality.

2. Too many voice attributes — More than 4 attributes dilute focus. Writers can't hold 7 attributes in mind while writing a tweet. Choose 3-4 that matter most and let the rest emerge naturally.

3. Generic do/don't examples — Writing abstract examples like "Be friendly, not formal" instead of showing actual sentences. Every rule needs a before/after pair using realistic brand copy.

4. Ignoring channel differences — A voice guide that treats Instagram captions and legal disclaimers the same way. The underlying personality stays constant, but tone and format must adapt to channel norms.

5. No vocabulary enforcement — Defining voice attributes without a word list. Writers will interpret "confident" differently; a vocabulary list makes it concrete and auditable.

6. Forgetting AI writers — Modern voice guides must include prompt-ready instructions, since a growing percentage of first drafts come from AI tools that need explicit voice constraints.

7. One-time document, never updated — Voice guides that get created during a rebrand and never touched again. Schedule quarterly reviews as the brand, audience, and channels evolve.

8. No scoring rubric — Without a way to measure voice compliance, feedback becomes subjective ("this doesn't feel right") rather than actionable ("this scores 2/5 on directness — here's why").

9. Skipping the audit step — Jumping straight to defining attributes without analyzing what the brand already sounds like. The audit reveals the voice that customers already recognize.

10. Making the guide too long to use — A 50-page brand bible that nobody reads. The best voice guides have a 1-page quick reference that writers pin to their wall, backed by deeper reference docs they consult when needed.

Resources#

• Output Template — Structured template for the complete brand voice guide deliverable
• Voice Attribute Framework — Deep-dive into selecting, scaling, and documenting voice attributes
• Channel Adaptation Guide — Framework for adapting voice across marketing, CX, product, and social channels
• Quality Checklist — Pre-delivery checklist covering completeness, consistency, and usability