cyberneticlibrary

Reverse-engineer and codify brand voice

brand-voice-analyzerskillsetup L1135
OneWave-AI/claude-skills

Causal-lift measurements

brand-voice-management-4pp vs no-skill baselinewith-skill 96% · baseline 100%

Measured by running the task with and without this artifact, K=5, graded by deterministic checks — no LLM judging.

What it does

Extract and codify a brand's voice from content samples across channels

Best for

When you need to codify how a company actually sounds so new writers can match the voice consistently.

Inputs
  • · Website copy, blog posts, emails, social media, sales collateral (3,000+ words minimum)
  • · Multiple content channels (2+ types)
Outputs
  • · Brand voice guide (markdown)
  • · Voice extraction across 7 tone dimensions
  • · Vocabulary inventory (power words, jargon, taboo terms)
  • · Signature phrases and sentence structures
Preconditions
  • · 3,000+ word corpus across 2+ channels exists
  • · Content is from real company output
Failure modes
  • · Corpus is too small (<1,000 words) — findings lack confidence
  • · Only analyzes one channel (incomplete picture)
  • · Confuses brand voice with business messaging
Trust signals
  • · Phase 1-4 workflow (collection → extraction → triage → review)
  • · 7-dimension tone spectrum (formality, humor, confidence, warmth, complexity, energy, irreverence)
  • · Vocabulary patterns (power words, filler, taboo, jargon, signatures)
  • · Sentence structure analysis with examples