Voice is a discipline, not a vibe — how we build tone-of-voice frameworks

Most voice documents are 30-slide decks listing adjectives. 'Confident.' 'Warm.' 'Honest.' 'Crisp.' Everyone nods. Nobody changes their writing.

Adjectives are not actionable. A copywriter sitting at 11 PM finishing a CTA doesn't refer to slide 17 to check whether 'Buy now' is 'confident enough'.

Sentence length range: e.g. 'mostly 8–14 words, occasional 3-word punches for impact'. Specific.

Vocabulary register: e.g. 'common-use English at 8th-grade level, no jargon unless naming a category we own'. Specific.

Idiom rules: e.g. 'no sports metaphors, no corporate jargon, no acronyms without first-use expansion'. Specific.

Punctuation conventions: e.g. 'em-dashes preferred over commas for emphasis, no semicolons in marketing copy'. Specific.

The framework is rules with exceptions. Like a typographic system. Predictable enough to be repeatable, flexible enough to feel human.

We map voice across two axes: formality (low to high) and energy (calm to charged). Every piece of copy lives somewhere on the grid. The brand has a NEUTRAL position (where most copy lives) and a RANGE (the deviation we allow on either side).

Crisis comms move toward formal-calm. Launch announcements move toward charged. Privacy policy is formal-calm-neutral. The grid keeps voice consistent without making it monotonous.

Show before/after. 'Before: We are pleased to announce... After: We just shipped...' Concrete examples teach the rule faster than the rule does.

Anti-examples are gold. 'Never write this' is more memorable than 'always write that'. We include 5–10 anti-examples in every voice doc.

Voice drifts. Every six months, audit recent copy against the framework. Find the 3–5 places voice has drifted. Tighten the rules or accept the drift consciously. Either is fine — drift unnoticed is the problem.