Brand Voice Guide
When five different freelancers write your product descriptions, a new CX rep drafts refund replies, and a TikTok creator records a UGC ad in the same week, your brand ends up speaking in five different voices — and customers feel the inconsistency even if they can't name it. Brand Voice Guide produces a concrete, example-driven voice document so anyone writing for the brand can match its sound on the first try, without needing a brand-team sign-off for every sentence.
Use when
- A founder says "I've been writing all our copy myself and now I'm hiring a copywriter — how do I hand off the brand voice without it sounding off-brand?"
- A team is expanding to new channels (TikTok, email, support, packaging inserts) and realizing the voice that works on Instagram doesn't translate to transactional emails or returns policies.
- A brand agency or creator collective asks for a written tone guide before starting a content batch, and the in-house team only has vague adjectives like "friendly but premium."
- A support team is rewriting canned responses and wants language that stays on-brand while handling refunds, shipping delays, and negative reviews.
What this skill does
Codifies brand voice along three dimensions: personality traits (3-4 specific adjectives with opposites, not just "friendly"), vocabulary (words and phrases the brand uses, words it avoids, industry jargon it translates for customers), and rhythm (sentence length, punctuation style, formality level, use of humor or emoji). It then adapts each dimension to the specific contexts where copy actually gets written — product titles, descriptions, email subject lines, push notifications, support replies, ad hooks, and packaging inserts — with side-by-side do/don't examples pulled from or inspired by the brand's existing best-performing content.
Inputs required
- Brand positioning statement (required): who you serve, what you sell, and the one thing that sets you apart; one or two sentences is enough.
- Personality adjectives (required): 3-5 words the founder or brand lead would use to describe the brand as if it were a person, plus any words they explicitly reject.
- Existing top copy samples (required): 5-10 pieces of copy the brand is proud of across different channels (product page, email, ad, social caption, support reply) to calibrate the voice from.
- Competitor or reference brands (required): two or three brands whose voice the team admires, and one or two whose voice they want to avoid sounding like.
- Channel scope (required): which surfaces need voice guidance (listings, email, SMS, support, ads, packaging, social); each gets its own adaptation block.
- Language and regional notes (optional): any regional phrases, spelling conventions (US vs UK), or translated-market considerations.
Output format
A structured voice guide organized into four parts. Part one is the voice foundation — personality adjectives with their opposites, one-line brand manifesto, and a "we sound like / we don't sound like" comparison. Part two is a vocabulary toolkit listing preferred words, banned words, signature phrases, and how to handle industry jargon. Part three is a channel-by-channel adaptation playbook: for each channel, a short description of the voice dial (more formal, more playful, etc.), three do/don't example pairs, and a template for the most common message type on that channel. Part four is an application checklist any writer can run through before publishing to confirm the copy sounds on-brand.
Scope
- Designed for: DTC founders, brand managers, content leads, and agencies creating voice documentation for ecommerce brands.
- Platform context: platform-agnostic, with adaptations for Shopify product pages, Amazon listings, TikTok Shop content, Klaviyo email, and support tooling.
- Language: English, with notes on regional variants.
Limitations
- Does not replace human judgment for edge cases like crisis communications or legal-sensitive copy.
- Voice documents drift if not maintained — expect to revisit annually or when the brand repositions.
- The guide is only as specific as the inputs; vague adjective inputs will produce a vague guide.
