Can AI write in our brand voice?

It can get much closer if you give it a concrete voice guide with real example passages, banned words and before-and-after rewrites. It will still drift toward generic phrasing and occasionally invent claims, so a person who knows the business should edit every piece before it is published.

What is the difference between brand voice and tone?

Voice is the consistent personality and point of view behind everything you write. Tone is how that voice adjusts to the situation, such as a celebratory announcement versus a reply to a complaint. Good guidelines define the voice once and describe how tone shifts.

How do I know if our brand voice is distinctive?

Cover your name and logo on a page of your copy and ask whether a competitor could have published it unchanged. If they could, the voice is not distinctive yet. Reading competitors’ websites aloud next to your own makes the shared phrases obvious.

Do we need a brand voice if we already have a logo and brand guidelines?

Visual identity tells people what you look like; voice tells them what you are like. Many guidelines treat voice as a single page of adjectives, which gives writers and AI tools nothing to work with. If your team writes from scratch every time, the voice section needs work.

Brand Voice in the AI Era: Staying Distinct When Everyone Writes With AI

Why everyone is starting to sound alike

Ask an AI writing tool for a homepage paragraph for an accounting firm and you get something competent: trusted advisors, tailored solutions, a commitment to your success. Ask again for a different accounting firm and you get nearly the same paragraph. That is not a flaw in the tool. Language models are built to produce likely text, and the likely text for an accounting firm homepage is the average of every accounting firm homepage.

When one firm in a market writes this way, it is forgettable. When most of them do, the whole category becomes interchangeable, and buyers fall back on price, proximity and whoever a friend mentioned. For an established business that has spent years building a reputation, that is an expensive way to save a few hours of writing.

The answer is not to ban the tools. It is to give them, and the people using them, something specific to aim at. That something is a voice, defined clearly enough that you can tell when a draft has missed it.

What a voice is actually made of

Most brand voice documents describe a personality in adjectives: friendly, professional, innovative, approachable. Every one of those words describes every competitor, and none of them helps a writer, human or machine, make a decision. A usable voice is a set of choices.

A point of view. What you believe about your field that not everyone does. A roofing company that thinks most roofs are replaced too early sounds different from one that thinks homeowners wait too long. Voice starts with having something to say.
Vocabulary. The words you use and the words you refuse. Do you say “clients”, “customers” or “members”? Do you allow “solutions”, “seamless”, “world-class”? A short banned list does more for distinctiveness than a long list of aspirations.
Rhythm. Short sentences or long ones, plain or ornate, how often you use a question, whether you ever use a fragment. Rhythm is what makes a paragraph recognizable before anyone reads the name.
Register by situation. Voice stays constant; tone moves. The same brand sounds different in a launch announcement, a reply to an angry review and an overdue invoice reminder. Spell out how.
Claims you will not make. No “best in Atlantic Canada” without proof. No promises about results you cannot control. Restraint is part of voice, and it is the part generic copy most often lacks.

Voice sits downstream of positioning. If you have not decided who you are for and what you claim, start with a positioning framework; voice cannot rescue a position that does not exist.

Finding the voice you already have

Established businesses usually have a voice already. It lives in the founder’s best emails, the way the senior partner explains a problem to a nervous client, the phrases customers repeat back in reviews. It rarely lives on the website.

Collect the raw material before writing a single guideline:

Your best real writing. Emails that won work, proposals that clients mentioned, the newsletter issue people replied to.
How your people talk. Record (with permission) a few sales calls or client explanations, or just ask your best people to explain what you do as they would to a friend.
Your customers’ words. Reviews, thank-you notes, the reasons people give for choosing you. These show which parts of your voice land.
Three competitors’ websites. Read them aloud. Note every phrase that also appears on yours.

The patterns that show up in the first three and not the fourth are your voice. The phrases that show up everywhere are what to remove.

Writing it down so people and tools can use it

A voice guide that will be pasted into an AI tool needs to work differently from one that sits in a brand book. Machines and busy staff both follow examples better than descriptions. Keep it short and concrete:

One paragraph describing the voice in plain language, including the point of view.
Four to six “this, not that” pairs, each with a real sentence on both sides. “Direct, not blunt” means little. A pair of example sentences shows exactly where the line is.
A word list: preferred terms, banned terms, and how you refer to your customers, your services and yourselves.
Five to ten exemplary passages from your best real writing, labelled by situation: web page, email, social post, reply to a complaint.
Two or three before-and-after rewrites showing a generic draft turned into your voice.
Bilingual notes if you publish in French and English. A voice does not translate word for word; decide how it carries into the second language, and have a fluent writer, not a tool alone, make that call.

Store this where the drafting happens: in the custom instructions or project settings of whatever AI tools your team uses, and in the shared folder where people start documents. For how this fits into broader guidelines that people actually open, see brand guidelines that teams actually use.

The editing loop

A voice guide makes AI drafts better. It does not make them finished. The step that protects your voice is human editing, done by someone who knows the business, against a short checklist:

Does this say something only we could say, or could a competitor sign it?
Has the draft added a cliché, a superlative or a claim we cannot prove?
Would our best salesperson say this out loud to a client?
Is every fact, name and number correct?
Does it sound right read aloud?

Name one owner for the voice. Not a committee, and not whoever happens to be using the AI tool that day. In a smaller organization this is often the marketing lead or a senior partner with a good ear; their job is to settle disputes quickly and keep the guide current, not to rewrite everything themselves.

When the same correction comes up repeatedly, add it to the guide. Over a few months the guide gets sharper and the drafts need less work. We cover where human review matters most in human review in AI marketing workflows.

A worked example

Consider an illustrative insurance brokerage in Saint John that serves commercial fishing operations and small manufacturers. Its AI-drafted services page opens like this:

We are a trusted partner committed to delivering tailored insurance solutions that protect what matters most to your business.

Nothing there is false, and nothing there is theirs. After a voice exercise, the brokerage settles on a point of view (most commercial policies are bought once and never read again, and that is where claims go wrong), a plain, slightly dry register, and a ban on “solutions” and “peace of mind”. The rewritten opening:

Most business owners read their commercial policy once, the day they sign it. We read it every year with you, because the boat, the building and the payroll have usually changed, and the policy usually has not.

The second version gives a buyer a reason to call, gives a competitor nothing to copy without looking foolish, and gives an AI tool a model to follow for the next twenty pages.

Does voice matter for search and AI answers?

Be careful with anyone who claims search engines or AI assistants reward a “distinctive voice” directly. That is not something the platforms describe. What they do describe is a preference for helpful, original content, whatever tool was used to produce it. And AI assistants quote passages that say something specific. Voice helps indirectly: a firm with a point of view tends to write content with substance, and substance is what gets read, cited and remembered.

Consistency helps too. When your website, your Business Profile description, your proposals and your social posts describe the business in the same terms, people and machines get one clear picture of who you are instead of four slightly different ones.

The bigger return is commercial. In a category where everyone sounds the same, the brand that sounds like itself is easier to remember, easier to refer and easier to choose. That is the work our brand strategy and identity engagements are built around: positioning, then the messaging that carries it, then a voice the whole team, and its tools, can hold to.