Your brand voice should be researched, not asserted

To research your brand voice, gather 12 to 20 candidate traits from your own copy and customer calls. Then run a MaxDiff study. Everyone who touches the brand picks the most and least fitting term across repeated sets. The ranked results show which traits win and where conflicts need a rule. Lock those rules so every output uses them.

One unresolved voice decision can split five assets: your website, sales deck, campaign, board update and AI draft.

For a fractional CMO managing four clients, that debate compounds an estimated 1.5 unbillable hours of daily context re-entry. Research the voice once, lock the decision and let every output use the same rules.

By the end of this article, you can run a six-step MaxDiff study that produces ranked voice traits, we-are/we-are-not pairs and messaging variants your team can apply.

Asserted brand voices fail at the first real tradeoff

Most brand voice workshops produce a list of agreeable adjectives. The company is visionary, dependable, practical, polished and inspirational.

Then someone has to write a sentence.

Can a visionary company sound pragmatic? Should polished copy become plain when the reader needs a direct answer? Does shipping quickly conflict with being predictable and dependable?

We saw these tensions surface during a working brand voice session. “Visionary” competed with “pragmatic.” “Polished” competed with “plain.” The useful decision was the priority between them.

That is where asserted voices break. A list describes the company everyone wants. It does not decide what a writer does when two desirable traits conflict.

A researched voice makes the tradeoff explicit before the website deadline forces it.

MaxDiff turns agreeable adjectives into ranked decisions

Maximum Difference Scaling, usually called MaxDiff, asks respondents to select the most and least fitting terms from repeated word sets.

The forced choice matters. Asking whether “inspirational” fits may produce five yes votes. Asking people to choose between inspirational, practical, visionary and dependable reveals relative preference.

Anonymous respondent links also change the meeting dynamic. The founder, junior marketer and outside CMO use the same tool. Each response carries the same weight.

The tradeoff is a little respondent effort upfront. The gain is a ranked signal that does not depend on who defended an adjective most forcefully.

How to research your brand voice with MaxDiff

1. Build a word set around real tensions

Start with 12 to 20 terms drawn from existing copy, leadership interviews and customer conversations.

Include words that create meaningful tension:

• Visionary versus pragmatic
• Polished versus plain
• Fast-moving versus dependable
• Opinionated versus prescriptive
• Encouraging versus hype-driven

Avoid easy opposites such as “clear versus confusing.” Nobody will choose confusing. The useful words force a decision between two defensible directions.

2. Include the people who see different parts of the brand

For a compact B2B team, invite five roles where available: founder, marketing, sales, customer success and delivery.

Send the same anonymous research link to each participant. Give them a deadline before scheduling a discussion.

Running the research first keeps the workshop from becoming the data source. The later conversation can explain the results rather than manufacture consensus.

3. Let repeated choices expose relative preference

Each respondent chooses the most fitting and least fitting term across repeated sets.

This is more useful than rating every word from one to five. Rating scales allow respondents to call every positive word important. MaxDiff requires priority.

Look for two signals in the results:

1. Consistent leaders: Terms that remain strong across respondents.
2. Useful disagreement: Terms that split the group and need an operating decision.

Disagreement is not failed research. It shows where future copy reviews would otherwise stall.

4. Resolve conflicts in the reader’s context

Take each disputed pair and test it against a real sentence.

If “visionary” outranks “practical,” that does not require abstract copy. It means the argument can look ahead while the supporting paragraph explains the mechanics.

If “dependable” outranks “fast-moving,” product announcements need evidence of control. Speed alone cannot carry the message.

Write down the deciding rule. That rule will do more work than another adjective.

5. Convert the winners into usable voice controls

Turn each selected trait into a we-are/we-are-not pair. Then add one behavior, one do and one don’t.

For example:

• We are: Visionary
• We are not: Vision-only
• This means: Show where the market is heading, then name the next operational step.
• Do: Connect a future claim to a mechanism or number.
• Don’t: Make broad predictions without showing how the reader acts.

Add mechanical controls where they matter. One team in our research named a direct rule: sentences should stay under 20 words. That instruction is easier to apply than “be concise.”

DISC variants adapt the message without changing the voice

A locked voice can still adjust to four common communication styles. Use DISC variants as editing modes, not personality diagnoses.

• Driver: Lead with the dollar cost and the decision that removes it.
• Influencer: Show the practice or company the reader is building.
• Steady Relator: Explain the incremental path and what remains intact.
• Conscientious Analyst: Give the mechanism, inputs and calculation before the claim.

The core argument stays fixed. The order and emphasis change for the reader.

Locking the voice turns research into an operating asset

In T2D3 OS, MaxDiff research, anonymous respondent links, tone rules and DISC variants live in the brand voice module.

Once approved, the voice can be locked as the version every downstream AI output reads. That removes the need to retype context into separate writing tools.

That re-entry cost adds up. Our estimate for a fractional with four clients is about 1.5 hours a day across 230 workdays. At $250 an hour, that is roughly $75,000 a year in unbillable time.

Locking the voice also immediately drafts the visual identity in the background. The visual work begins from an approved verbal direction rather than another disconnected workshop.

Start with one client or one internal brand. Run the research, resolve the top conflicts and test the locked voice across five assets. Review exceptions as evidence for the next version, rather than reopening every settled decision.