Brand Voice & Tone
Brand Voice and Tone
What is brand voice?
- Voice stays constant; tone adjusts to the situation
- Captured as do/don’t sentence pairs, not adjectives
- It works only when built into the templates of departments outside marketing
Brand voice isn’t about picking a few adjectives like “friendly and professional.” It’s about creating a standard that makes anything anyone writes read as coming from the same company. If brand identity defines the “look” through logo, color, and typeface, voice defines the “sound.” Voice stays constant like a person’s character, while tone shifts with the situation, much as you speak differently at a funeral and at a celebration. An outage apology and a product announcement can differ in tone, yet they should read as if the same person wrote them.
B2B companies need this standard because so many people write. Marketing writes the website, sales the proposals, the technical team the manuals, HR the job postings, and customer support the notification emails. Customers read all of it as one company’s voice. To be usable, a guide should be built not from adjectives but from pairs of 'write this / don't write this' sentences. 'Spell out abbreviations on first use' changes real writing more than 'professional but not difficult' does.
Four things are enough to include: three or four voice traits, each with a limit (speak with confidence, but don’t exaggerate); a tone-adjustment table by situation (announcements, apologies, congratulations, sales); terminology and style rules (forms of address, honorifics, numbers and units, banned words); and about ten before-and-after rewrite examples. Keep it to two or three pages. For organizations that hand first drafts to generative AI, this document becomes both the standard for prompts and the review criteria for screening the output.
The usual breakdown in practice is a guide that lives only in the marketing team’s drawer. Sales proposals, job postings, and customer support notices keep the old tone, so the voice customers hear is still inconsistent. The opposite happens too: rules that are too granular. With hundreds of banned words, nobody reads them, and every piece is forced into one mold and loses its life. Nor is it a one-time document. It stays a living standard only if, once a quarter, the team reads a few published pieces together and checks them against it.
John KimInstructor John Kim, author of "B2B Marketing Design" and CEO of Nanum Consulting
Jongsung EunExpert in AI, marketing, business models, and management strategyA concept that became part of brand guidelines as brand identity management expanded from visual elements to verbal ones. It has no single originator. It was refined in advertising, editorial, and UX writing practice, and companies call it by different names, such as 'voice and tone' or 'verbal identity.'
Consider a hypothetical case. Company A, which makes enterprise accounting software, has 180 employees. A customer survey produced 31 responses saying 'the website is friendly, but outage notices and contract emails feel like a different company,' and job postings used different forms of address and writing styles from team to team.
- The marketing lead collected 40 pieces sent to customers over the past three months. The terms used for customers alone came in four variants: 'customer,' 'your company,' 'manager,' and 'user.'
- One person each from marketing, sales, customer service, and HR met and narrowed the voice to three traits: “clear, courteous, plain-spoken.” They also added a one-line limit for each that must not be crossed.
- The customer service team lead said, “Adjectives won’t work for us. Give us examples.” The team added 12 before-and-after examples, rewritten from real outage notices and contract guides.
- They built the two-page guide into proposal templates, email formats, and job-posting templates. They also put the same rules and 15 banned expressions into the default instructions of the AI drafting tool.
- Four months later, the same survey showed “feels like a different company” responses falling from 31 to 6. Internal review time for outage notices also dropped, from an average of 50 minutes to 20.