Brand Voice Guide

SKILL.md

---
name: Brand Voice Guide
description: Define and document a consistent brand voice with tone guidelines, vocabulary lists, do/don't examples, and channel-specific adaptations so every touchpoint sounds unmistakably like the brand — whether written by founders, freelancers, or AI.
---

# 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

| Decision | Strong | Acceptable | Weak |
|---|---|---|---|
| Voice attributes | 3-4 defined with spectrum scales, each with do/don't examples | 3-4 attributes listed with brief descriptions | Vague terms like "professional" with no examples |
| Vocabulary list | 50+ terms with approved/banned alternatives and context notes | 20-30 approved/banned terms | No vocabulary guidance or just a few random words |
| Tone spectrum | 5+ scenarios mapped on a casual-to-formal scale with examples | General guidance on when to be formal vs. casual | Single tone for all contexts |
| Channel adaptation | Separate guidelines per channel (email, social, ads, CX, product) with examples | General channel notes without examples | One voice doc applied identically everywhere |
| Do/don't examples | 10+ paired examples per channel showing exact rewrites | 5-10 generic do/don't pairs | Abstract rules with no concrete examples |
| Audience awareness | Voice adjusted by segment (new vs. loyal, B2B vs. B2C) | Single audience definition | No audience consideration |
| Testing method | Before/after audit of real brand copy with scoring rubric | Informal review by brand team | No validation process |
| AI prompt integration | Voice attributes formatted as system prompt instructions | General AI usage notes | No 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:

| Context | Tone Shift | Example |
|---|---|---|
| Product launch announcement | Peak excitement, confident | |
| Error message / outage notification | Calm, direct, empathetic | |
| Customer complaint response | Warm, accountable, solution-focused | |
| Sales email | Enthusiastic but not pushy | |
| Legal / compliance | Clear, formal, still human | |
| Social media comment reply | Casual, quick, personality-forward | |
| Onboarding sequence | Encouraging, helpful, patient | |
| Win-back / re-engagement | Personal, 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:**
| Approved | Banned | Context |
|---|---|---|
| "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:**
```
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](references/output-template.md) — Structured template for the complete brand voice guide deliverable
- [Voice Attribute Framework](references/voice-attribute-framework.md) — Deep-dive into selecting, scaling, and documenting voice attributes
- [Channel Adaptation Guide](references/channel-adaptation-guide.md) — Framework for adapting voice across marketing, CX, product, and social channels
- [Quality Checklist](assets/quality-checklist.md) — Pre-delivery checklist covering completeness, consistency, and usability