Tone of Voice

What it is

How a brand writes — observed, not imagined

TONE.md is extracted from the actual copy on a company's website. It captures the formality level, sentence construction habits, vocabulary choices, and emotional register the brand uses consistently — not the voice it claims to have in a PDF guidelines document. The output is structured for direct use in AI writing pipelines.

Output

What the file contains

Voice adjectives

3–6 adjectives that characterize how the brand writes.

Formality level

Calibrated register: formal, professional, conversational, or casual.

Sentence patterns

Typical length, complexity, and construction style (declarative, imperative, interrogative).

Vocabulary profile

Words the brand favors and words it avoids — with examples from source copy.

POV and stance

First-person, second-person, or third-person preference; proximity to the reader.

Emotional register

Where the brand sits on the warm–cool and serious–playful spectrum.

How it works

Pattern extraction, not guesswork

1. 01

   Copy sampling

   The analyzer reads homepage headlines, subheadings, body copy, button labels, and blog excerpts — pages where voice is most deliberately crafted.

2. 02

   Pattern analysis

   A language model identifies formality markers, sentence construction patterns, recurring vocabulary, and stance (first-person, second-person, authoritative, collaborative).

3. 03

   Structured output

   Findings are written as a TONE.md file with named dimensions, example phrases lifted from the source, and usage guidance. Cached by domain.

FAQ

Common questions

What is a tone-of-voice analysis?

A tone-of-voice analysis identifies how a company writes — the adjectives that describe the brand's voice, the formality level, the sentence structure patterns, and the vocabulary the brand consistently uses or avoids. The output is a structured reference that can be applied directly to content production.

How is TONE.md different from a brand guidelines document?

Brand guidelines are written by the company for human designers. TONE.md is generated from observed writing patterns on the live site — it reflects how the brand actually sounds, not how it intends to sound. The two often differ. TONE.md is also formatted for direct use in AI content pipelines.

What does the analyzer read?

The analyzer reads the homepage and key content pages — about, blog posts, and product descriptions. It extracts patterns from real copy: sentence length, formality markers, first-person vs. second-person stance, imperative vs. declarative constructions, and recurring vocabulary clusters.

How is TONE.md useful for AI content generation?

When generating content with an LLM, providing a TONE.md file as context dramatically reduces generic output. The model writes in the brand's actual register — matching sentence length, vocabulary level, and emotional tone — rather than defaulting to a generic professional voice.

Is the analysis free?

Yes. The public tone-of-voice analyzer is free with no account required. Results are cached per domain, so if the same site has already been analyzed, the result returns immediately.