Brand Voice Doc Builder

---
name: brand-voice-doc-builder
description: Use when building a brand voice and tone guide from sample writing. Triggers on "brand voice doc", "voice guide", "tone guide", "style guide for content".
---

# Brand Voice Doc Builder

Most voice docs are useless because they list adjectives ("smart, approachable, bold") without examples. A writer reading "approachable" can't actually write differently. A good voice doc shows them.

## Required inputs

1. **3-5 samples** of your best existing writing — blog, emails, founder talks, customer comms
2. **3-5 samples** of writing that's NOT your voice — corporate-speak, competitor copy, off-brand drafts
3. **Audience** — primary reader + their context
4. **Brand position** — challenger / category leader / specialist / generalist
5. **What's changed recently** — rebrand, repositioning, new persona

If the user provides no samples, push back. You can't extract a voice from a vibe.

## Extraction process

### Step 1: Read everything in the "yes" pile
Find the patterns. Note:
- Sentence length distribution (mostly short? long? mix?)
- First-person vs second-person vs third-person
- Use of contractions ("don't" vs "do not")
- Jargon density
- Concrete nouns vs abstract concepts
- How metaphors and analogies are used
- Where humor / wit shows up (or doesn't)
- How they handle uncertainty ("we think" vs "we know")

### Step 2: Read the "no" pile
Note what they're explicitly NOT. This is often more useful than what they are.

### Step 3: Extract 4-6 voice attributes
NOT "smart, approachable, bold." Specific behaviors. Examples:
- "Direct without being curt"
- "Specific over generic — uses real numbers, not 'significant growth'"
- "Confident, not boastful — 'this works' not 'this is the best'"
- "Skeptical of jargon, but uses precise technical terms when it matters"

Each attribute needs a worked example.

## Output structure

```
# [Brand] Voice Guide

## Who we're writing for
1-2 paragraphs: the reader, in detail.

## How we sound (4-6 attributes)
For each:
- Attribute name
- 1-2 sentence definition
- Do example (from real writing)
- Don't example (corporate-speak version)

## Vocabulary
### Words we use
[10-15 words/phrases that recur in the brand]
### Words we don't use
[10-15 corporate-speak / competitor-speak we avoid]

## Tone calibration by context
| Context | Tone shift |
|---------|------------|
| Sales email | More direct, less playful |
| Blog | Educational, longer-form |
| Social | Punchy, observational |
| Support reply | Patient, no jargon |
| Internal | Candid, less polished |

## Punctuation and style
- Oxford comma: yes / no
- Em dash: liberal / sparing
- Capitalization style for product names
- Number formatting (3 vs three; 1,000 vs 1000)
- One-sentence paragraphs: allowed / discouraged

## Worked examples
Take a flat corporate sentence. Show the 3 voice-aligned rewrites.

Example:
- BEFORE: "Our solution empowers teams to streamline their workflow."
- AFTER (this brand): "We make Mondays less painful for ops teams. That's it."

Provide 5 of these.

## Anti-patterns specific to this brand
[5-10 phrases or patterns that ALWAYS read as off-brand. Cite where they came from in the no-pile.]
```

## Anti-patterns in voice docs themselves

Strip these patterns from any draft:
- Adjective lists with no examples
- "Bold but humble, smart but accessible" — vague spectrums
- Voice attributes that are actually values ("we believe in honesty")
- Generic "do / don't" that any brand could claim

## Tone

- The voice doc itself should be written in the voice it describes
- Write it for a contractor on day one, not a brand strategist
- One page if possible. Two max. Long voice docs don't get read.

## Output

Markdown. After the doc, list 3 places where the user's samples disagreed with each other — the voice has internal tension worth resolving.