Bloom is now Rainbrand

By Siddharth Borderwala

Published

11 min read

Brand voice vs tone: what's the difference?

Brand voice is the personality that stays the same; tone is how it adjusts to the situation. See the difference in real brands' copy, build a tone map for the situations you write for, and write tone rules an AI agent can follow.

TL;DR

Brand voice is how a brand sounds every time: its personality in words. Tone is how that voice adjusts to the situation and the reader's state, from a launch to a payment problem. Write the voice as a few traits with limits and examples. Write tone as a map of the situations you write for, with what changes, what never does and an approved example of each.

Venmo, a US app for paying friends back, calls itself "an app where settling up feels like catching up" on its About page, with "bajillions of delightful payments (and emojis, and inside jokes)". Its help article on payment holds says: "I know it can be concerning when money is on hold for a payment you've received, but remember that having a payment on hold doesn't mean that you've done anything wrong."

Both sound like Venmo. They use plain words, talk to one person and skip the jargon. That's the voice. One is having fun and the other is calming someone who's worried about their money. That's tone.

Voice, tone, personality and style

The four terms overlap, and guides use them differently. Here's how they usually divide:

TermWhat it isDoes it change?
PersonalityThe character behind the brand, as a few traitsNo
VoiceHow that personality sounds in wordsNo
ToneHow the voice adjusts to the situation, the reader's state and the stakesYes, by situation
StyleHouse rules for spelling, capitals, numbers and punctuationRarely

Messaging is a fifth thing that gets mixed in: it's what you say, such as claims and proof, rather than how you say it. Style and the rest of a brand's rules sit in its brand style guide.

Where guides disagree is mostly about which term holds the others. Frontify treats voice as an umbrella with tone and personality underneath it. Adobe's design system keeps them side by side: "We speak with one voice and vary our tone depending on situational context." And the vocabulary varies by country. Nick Parker, a verbal identity consultant, notes that "tone of voice" became the usual term in the UK for the whole thing, while "brand voice" is more usual in the US, where tech companies in particular split voice from tone.

The split is worth keeping whatever you call it, because the two are written and checked differently. A voice rule applies to every message, so it goes in every brief. A tone rule applies to one situation, so whoever is writing, a person or an agent, has to know which situation they're in. If your team says "tone of voice" for the whole thing, keep the name and still separate the two inside the guide.

Write the voice as traits with limits

A voice is usually three or four traits. On their own, adjectives such as "friendly" or "bold" don't tell a writer much, because every brand wants to be friendly. What makes a trait usable is its limit: how far it goes before it turns into something you don't want.

PostHog, which makes product analytics software, puts this in a table in its brand handbook:

Feel like thisNot like this
OpinionatedDiplomatic to the point of saying nothing
HumanCorporate robot
Slightly weirdTrying to be funny in a try-hard way
ThoughtfulRandom
DirectFluffy
HonestCorporate fluff
PlayfulChildish or unprofessional
ApproachableArrogant

The right-hand column is what makes it usable. "Playful" alone could justify almost anything; "not childish or unprofessional" tells a writer where to stop. The same handbook shows the voice in a sentence: "We don't want to be another boring B2B SaaS company."

Then add real lines that show the traits working. Voices that are well defined stop fitting when you swap them between brands:

  • Linear, project tracking software, is restrained and practical. Its careers page lists principles such as "Simplify and ship smaller" and "Get to the point and assume good intentions."
  • Headspace, a meditation app, is warm and reassuring: "There's no such thing as perfect meditation," says its guide to meditation.
  • PostHog's line above would sound wrong from either of them, and theirs would sound flat from PostHog.

Brand voice examples reads seven published voice guides in depth, and how to write brand voice guidelines AI tools can follow covers building the full guide from real edits.

The four tone dials

Tone is easier to set when you name what's changing. The Nielsen Norman Group, a user-experience research firm, describes four dimensions of tone, each a scale:

  • formal to casual;
  • serious to funny;
  • respectful to irreverent;
  • matter-of-fact to enthusiastic.

In its study, 50 people in the US rated versions of the same copy that differed only in tone. A funny, casual sample for an insurance company came across as friendlier but less formal than a serious one. The differences were small, about half a point to a point on a five-point scale, but they were statistically significant, and the article's advice is to "vary the tone to fit the user's emotional state and the topic."

Treat the four as dials you set per situation, inside the limits your voice allows. If your voice rules out irreverence, that dial stays put everywhere.

Map tone to situations

A tone map lists the situations you write for and sets the dials for each. Adobe's Spectrum design system has one of the clearest published versions. It places tone on a spectrum:

A horizontal gradient bar with five points labelled, from left to right: Motivational, Helpful, Instructive, Reassuring and Supportive

Adobe Spectrum's tone spectrum

Then it gives each point an attitude, a line that captures it, and how often to use it:

Tone and attitudeA line in that toneHow often
Motivational: positive and encouraging"We're looking out for you and cheering you on. You've got this!"Rarely
Helpful: polite and respectful"We know you're busy, so we'll make this brief."Occasionally
Instructive: neutral and direct"Here is the information you need."Often
Reassuring: professional and reliable"We know you're worried about this issue, and we're here to help."Occasionally
Supportive: concerned and empathetic"Something bad has happened and we understand how you feel."Rarely

The frequency column is the unusual part. It tells writers that most product copy should be plain instruction, and that cheering someone on or consoling them are for the moments that call for it. A writer who reaches for "You've got this!" on every screen has the right tone in the wrong place.

Intuit's content design system makes the same kind of call for notifications: "more emotional tone for milestones and celebrations" and "more functional tone for alerts and urgent to-do's." Its example of the first: "Great news! Your federal tax return was accepted by the IRS."

Build your own map from the situations you actually write for. One entry per situation is enough:

Situation:         [a customer's payment failed]
Reader's state:    [mid-task, maybe worried]
Formal ↔ casual:   [where it sits]
Serious ↔ funny:
Respectful ↔ irreverent:
Matter-of-fact ↔ enthusiastic:
Always:            [plain words, the next step]
Never here:        [jokes, blame, promised dates]
Must carry:        [amount, payee, the money's status]
Approved example:  [one real message]

One message, three tones

Here is one message, written at three settings of the dials. The situation: a card payment was declined, so no money moved. These are written for this article, for an unnamed payments app, to show the dials moving:

DialsThe message
Formal, serious, respectful, matter-of-fact"Your payment of $40.00 to Sam Lee could not be completed. No funds have been withdrawn from your account. Please verify your card details and try again."
Casual, serious, respectful, matter-of-fact"Your $40 payment to Sam didn't go through, and no money left your account. Check your card details, then try again."
Casual, funny, irreverent, enthusiastic"Well, that bounced! Your $40 to Sam didn't make it, but hey, your money's still yours. Give your card details a once-over and fire it off again!"

All three are clear about what happened. The first makes the reader work harder for it: "funds", "withdrawn" and "verify" are form-letter words, and nothing in it acknowledges that a declined payment is an annoying moment. The second says the same in plainer words and leads with what the reader is anxious to know. The third makes light of a problem with someone's money, and Adobe's frequency column explains why that fails: a worried reader needs reassurance or plain instruction, not a joke.

Notice what stays the same in all three: the amount, the person, what happened to the money and the next step. Those are facts and structure, and they belong in every version. The dials only change how it sounds.

Tone rules an agent can follow

An agent writing a message doesn't know the situation unless the task says so. Given "write a notification about a failed payment", it has to guess how the reader feels, and the guess sets the tone.

Rainbrand maintains brand guidelines and files as context that AI agents call while they work. Its brand guidelines library has a page for Venmo, built from Venmo's public site, and the same brand can be asked for a Brief: the guidance and files from the brand that one task needs. Two Briefs were requested for Venmo, one on November 15, 2026 and one on November 16. One task was an Instagram caption inviting friends to split the bill after a birthday dinner. The other was an in-app message telling a customer that $40 they received from a friend is on hold; that task supplied the facts from Venmo's help article on holds, because the Brief has no way to know them. The voice came back the same, and the tone didn't:

Birthday captionPayment on hold
Voice"short, friendly, direct, and conversational""friendly, direct, and conversational"
Humour"A light touch of humor or personality can work, but keep it genuine""use a calm, empathetic light touch rather than playful banter"
Lead with"Connect splitting the bill with the shared celebration""Clearly say that the $40 payment from their friend is on hold"
Avoid"stiff, impersonal, or finance-heavy wording"inventing "a review duration" or promising "an outcome beyond that condition"

The hold Brief also drew a line between the two kinds of input. Its voice guidance came from the brand; the amount, the steps in the app and what happens after the review were "facts supplied in the task", not brand rules.

Here is the copy written from each Brief, by Claude, on November 16, 2026, using only the Brief and the task:

Birthday caption. Birthday dinner done, bill on the table. Split it on Venmo and get back to the cake.

Payment on hold. The $40 your friend sent you is on hold. Holds are a common way to help keep everyone safe, and this one doesn't mean you did anything wrong. To see why it's on hold and whether you need to do anything, go to the Me tab, tap Transactions and open the payment marked On-hold. Once the review is complete with no problems, the money will be in your account.

What stayed the same is the voice. Both use everyday words, talk to one person as "you", keep sentences short and avoid finance words such as "funds" or "authorization". What changed is the tone:

The tone decisionBirthday captionPayment on hold
First thing it saysThe scene: dinner's overThe status: the money is on hold
HumourA small one: get back to the cakeNone
ReassuranceNot needed"doesn't mean you did anything wrong"
What it asks of youSplit the billCheck the payment in three exact steps
What it promisesNothingOnly Venmo's own condition, a review with no problems

Neither message has been tested on Venmo customers, and the caption's joke is the kind of choice a person should approve. But you can see each dial move, and you can see which words the brand fixed and which the situation chose.

That run shows what an agent needs from your tone rules:

  • The situation and the reader's state, named in the task. "A customer whose payment is on hold" sets the dials; "a notification" doesn't.
  • A "never" for each situation. No jokes about money problems, no blame, no dates you can't promise. A limit written down is one the agent doesn't have to infer.
  • Facts kept apart from tone. The tone rule says to be reassuring; only the task can say when the review ends, if anyone knows.
  • One approved message per situation. An example shows where the dials sit faster than a description.

Keeping AI writing in your brand voice across tools covers how to give every agent the same rules. Then read what comes back by situation, not by tool: put five payment-problem messages side by side and a joke stands out in a second.

You don't need a map of every situation before you start. Write entries for the three you write most often and the one that goes wrong most often, and add the rest when a message lands in the wrong tone.

FAQ

Does tone change by channel or by situation?

Mostly by situation. A complaint answered on Instagram and one answered by email need the same care, even if the format differs. Channels matter for format and length; the reader's situation sets the tone. Brand Vision's guide makes the same point: keep channel guidance "subordinate to the situation".

How many voice traits should a brand have?

Three or four is common, and few enough to remember. Adobe uses three (rational, human, focused). PostHog lists eight, but each comes with its limit, which keeps a longer list usable. More traits than writers can recall stop shaping the writing.

Can a brand have more than one voice?

A company with several brands usually gives each its own. Intuit's content design system has separate voice pages for Intuit, Mailchimp, QuickBooks and TurboTax. Within one brand, keep one voice and let tone do the adjusting; two voices for one brand usually means the guide hasn't decided which it is.

Where does humour belong, in the voice or the tone?

Both. The voice decides whether humour is allowed at all: Adobe's voice tells writers to avoid language that's "overly opinionated, funny, or trendy", while PostHog's asks for "slightly weird". The tone decides where it's allowed, which is rarely in errors, bad news or anything involving money.

All 5 Brand voice articles
  • 14 min read

    Brand messaging framework: a template, a completed example and how to test it

    How to build a brand messaging framework from customer research. Download a free template, see it completed for GitLab from public sources, read the ad, email and page copy an AI agent wrote from it, and learn how to test the messages with your audience.

  • 5 min read

    How to keep AI writing in your brand voice across tools

    ChatGPT, Claude, Gemini, Jasper and your own agents each bring their own defaults, settings and copy of your guide, so one voice comes out several ways. Write the voice in the parts that carry across models, keep one version every tool reads, and check the same tasks in each tool.

  • 11 min read

    Brand voice examples, and how to write rules AI can follow

    Brand voice examples from Slack, Innocent, Apple, Visa, Oatly, Nike and eight published voice guides, with the real copy behind each and rules your AI tools can follow.

Connect Rainbrand.

  1. Open the Rainbrand plugin and add it.

    Open Rainbrand plugin (opens in a new tab)
  2. Connect your Rainbrand account when asked. It works in ChatGPT and in Codex.