Brand Voice Skill

Overview

This skill provides a systematic approach to brand voice application: extracting voice guidelines from the RAG knowledge base, calibrating tone for specific platforms and audiences, and auditing content for brand consistency.

Tools to Use

• RAG (brand_guidelines): Primary source for brand voice, values, and messaging
• Perplexity: Research how the brand category typically communicates; identify differentiation
• Jina: Analyze existing brand content on website/social for voice audit

Brand Voice Components

The 4 Dimensions of Brand Voice

1. Tone — How does the brand "feel"?

   • Spectrum: Formal ←→ Conversational
   • Spectrum: Serious ←→ Playful
   • Spectrum: Authoritative ←→ Humble
   • Spectrum: Inspirational ←→ Practical

2. Vocabulary — What words does the brand use/avoid?

   • Brand-specific terms and phrases
   • Words to actively use (power words)
   • Words to avoid (corporate-speak, clichés)
   • Technical vocabulary level

3. Rhythm — How does the brand structure sentences?

   • Short, punchy sentences vs. flowing prose
   • Use of lists vs. narrative paragraphs
   • Punctuation style (minimal vs. expressive)
   • Cadence (fast/urgent vs. measured/thoughtful)

4. Persona — Who is the brand in the reader's mind?

   • Role archetype (Mentor? Peer? Expert? Guide?)
   • Level of familiarity (use "you" / "we" / formal)
   • Cultural references and humor level

Workflow

Step 1 — Load Brand Voice from RAG

Query the brand_guidelines RAG source for:

• Brand personality descriptors (3–5 adjectives)
• Explicit do's and don'ts for copy
• Example copy that exemplifies the brand voice
• Core values that should permeate all communication

Step 2 — Platform Calibration

Brand voice is constant, but tone adjusts per platform:

Step 3 — Audience Calibration

Adjust vocabulary and complexity for:

• C-suite: Strategic, outcome-focused, minimal jargon
• Practitioners/Professionals: Specific, technical where earned, peer-to-peer
• General audience: Plain language, relatable examples, no industry terms

Step 4 — Voice Audit (for existing content)

When reviewing content for brand consistency, check:

Tone checklist:

• Does the opening line sound like the brand?
• Is the formality level consistent throughout?
• Are there any phrases that feel "off-brand"?
• Does it end with the right emotional note?

Vocabulary audit:

• No off-brand words used
• Brand's preferred terminology applied correctly
• No jargon unless appropriate for audience
• Consistent pronoun use (you/we/our)

Rhythm check:

• Sentence length variation appropriate to brand
• Paragraph breaks align with brand style
• Lists vs. prose ratio correct for platform

Step 5 — Voice Calibration for Drafts

Before writing new content:

1. Identify the 3 brand voice adjectives most relevant to THIS piece
2. Write a "voice reminder" sentence (e.g., "Write as if you're a trusted mentor who is direct, warm, and backed by data")
3. After drafting: read aloud — does it sound like the brand?

Step 6 — Voice Consistency Score

Rate drafted content 1–5 on each dimension:

• Tone alignment: [1–5]
• Vocabulary fit: [1–5]
• Rhythm match: [1–5]
• Persona consistency: [1–5]

Minimum threshold: 4+ on all dimensions before approving content.

Output Format

## Brand Voice Brief

**Platform**: [LinkedIn / Instagram / Blog / etc.]
**Audience**: [Target reader profile]
**Content goal**: [Awareness / Engagement / Conversion]

### Active Voice Profile for This Piece
- **Tone**: [3 adjectives from brand voice]
- **Formality**: [1–5, where 1=very casual, 5=very formal]
- **Persona**: [Brand as: Mentor / Peer / Expert / etc.]
- **Key vocabulary**: [Words to use prominently]
- **Words to avoid**: [Off-brand terms]
- **Rhythm**: [Short & punchy / Flowing narrative / Mixed]

### Voice Audit (if reviewing existing content)
- Tone alignment: [1–5] — [notes]
- Vocabulary fit: [1–5] — [notes]
- Rhythm match: [1–5] — [notes]
- Persona consistency: [1–5] — [notes]
- **Recommended edits**: [specific changes]

### Voice Calibration Sentence
"[Write this piece as if you are a [persona] who is [adjectives]...]"

Notes

• Always query brand_guidelines RAG first — do not invent a voice profile
• When brand_guidelines are sparse: use Jina to extract voice from existing brand content on website/social
• For multi-brand projects: create separate voice profiles per brand; never blend
• Italian language brands: pay attention to cultural register (formal "Lei" vs. informal "tu") as this is a brand decision, not just grammar