Amy Baker.
All case studies

One voice for a business where everyone writes

At a B2B technology company, I led a group from across the business to agree how we wanted to sound, and turned it into a tone of voice guide. I then built an AI prompt that checked any draft against the guide and explained each fix. Around 130 pieces of content went through it.

Content checked against one voice

Plus sales collateral

~130Pieces checked
~50
Website pages
~50
Blog posts
~30
Case studies

Challenge

Sales, leadership, people and culture, and marketing all wrote for the business, and each sounded different.

What I did

Led a tone of voice with a small cross-team group, then built an AI prompt that checked any draft against our rules and explained each fix.

Result

Around 130 pieces, including about 50 blogs, 30 case studies, and 50 website pages, brought into one voice without starting again from scratch.

The challenge

A lot of people wrote for the business. Sales built slide decks. Our head of people and culture wrote hiring and onboarding content. The CEO wrote to clients. I wrote blog posts and case studies. Each person wrote well, but in their own way, so a buyer could read three pieces from us and hear three different companies.

We had just agreed our positioning, values, and what made us different. We’d also mapped our target personas, the customer journey, and our marketing, sales, and account management pipeline. The way we wrote didn’t yet reflect any of it. We wanted to sound like a thoughtful peer to our clients, not an authority telling them what to do, in plain and human language.

Lucidpress’s 2021 Content Effectiveness Report found 77% of professionals see off-brand content created at their company. With this many writers and no shared rules, we were likely to be one of them.

Source: Lucidpress, “2021 Content Effectiveness Report”.

My approach

I put together a group of four, with one person each from people and culture, design, the technical team, and me for marketing and sales. The team had already done a thinking-styles profile, and each of us had a different style, so the group covered different ways of reading and writing as well as different parts of the business.

We started with real content. Each of us collected a case study, a blog post, a piece of sales collateral, and a client communication, and marked where we felt it didn’t sound like us. We also found pieces we thought did sound right. Over a few meetings we compared our comments, rewrote sections, and worked out where we agreed and where we didn’t. Every decision went back to our values, our positioning, and what our personas needed to hear.

What I built

A tone of voice guide. A PowerPoint guide covering how we sound, how we don’t sound, and the rules for writing as the business, with examples of each. The rules included:

  • Keep sentences short. Our readers were business owners and the C-suite, without time for long explanations.
  • Back up every claim. Use research or examples of our own work, or leave the claim out.
  • No hyperbole. Keep it grounded in what actually happened.
  • Don’t over-promise.
  • Explain, don’t instruct. Talk as a peer, without telling people what they should do.
  • No AI-sounding filler. No sweeping openers like “in today’s world” or “in times like these”.
Tone of voice guideSlide 4 of 12
How we sound
“Here’s what we found when we tested it with three clients.”
“This worked for us. It might not suit every team.”
“We cut onboarding from six steps to three.”
How we don’t sound
“In today’s fast-paced world, every business needs…”
“The only partner you’ll ever need.”
“You must act now or fall behind.”
Short sentencesBack up every claimNo hyperboleDon’t over-promiseExplain, don’t instructNo AI filler
How the guide was set out, with example lines written for this page: how we sound, how we don’t, and the rules.

An AI tone of voice checker. A prompt the team pasted into Gemini with any draft. It held every rule from the guide and asked the AI to:

  1. Rate the content against the guide.
  2. Highlight each place a rule was broken and say which rule.
  3. Explain why it broke the rule.
  4. Recommend how to improve it.
Tone of voice checkRating: 6 out of 10
Example draftIn today’s fast-paced world,1 healthcare providers need a partner they can trust. Our revolutionary2 approach guarantees results for every client3.
1No AI filler
Why: A sweeping opener that says nothing about the reader.Try: Start with their problem.
2No hyperbole
Why: A big claim the rest of the page can’t prove.Try: Say what the approach does.
3Don’t over-promise
Why: We can’t guarantee results for every client.Try: Point to a real example of our work.
An example of the checker’s feedback on a made-up draft: a rating, each rule broken, why, and what to try.

I wrote it to teach as well as fix. Each explanation showed the writer the pattern, so over time they could write in the voice from the start without needing the checker.

A lighter check for personal writing. Emails and blog posts written under a person’s name kept their personality. For those, the checker was a sense check against the core rules, not a full rewrite into the company voice.

Results

Where the checker applied

Company voice where we spoke as a business, a lighter check where a person did

AI tone of voice check
Full check, company voice
Website pagesCase studiesSales collateral
Sense check, own voice
EmailsNamed blog posts
  • Around 130 pieces checked. About 50 blogs, 30 case studies, and 50 website pages, plus our sales collateral, went through the checker and were updated with its fixes.
  • No full rewrite. Existing content was brought into line with the new voice and messaging, with nothing removed or written again from scratch.
  • One voice across writers. It became hard to tell which person had written which page, while named blogs and emails still sounded like their authors.
  • Used across the team. The team ran it on their own drafts. I also ran it on content other people wrote and made the changes.

Want your whole team to sound like one business?

I’ll look at how your content reads today.

Get in touch

What I’d do next

I’d set the checker up as a shared custom assistant, such as a Gemini Gem, with the guide attached. That removes the copy-and-paste step, and every update to the rules reaches everyone at once.

I’d also keep a log of which rules the checker flagged most. That would show where people needed more help, and which rules needed clearer examples in the guide.

FAQs

How do you create a tone of voice for a B2B business?

Start with your positioning, values, and personas, then look at your real content. Ask a small group from different teams to mark what does and doesn’t sound like you, and turn the points you agree on into rules with examples.

Can AI check content against your tone of voice?

Yes, if you give it clear rules. A prompt that lists each rule and asks the AI to highlight the problem, name the rule, explain it, and suggest a fix works with ChatGPT, Gemini, or Claude.

Won’t an AI checker make everyone sound the same?

It can, so decide where the company voice applies. We used the full check on website pages, case studies, and sales collateral, and a lighter check on emails and named blog posts so they kept their author’s personality.

Let's talk about your marketing.

Tell me a bit about your business and what you need, and I'll take it from there.

What happens next

  1. 1You tell me about your business and what you need.
  2. 2I reply within one business day.
  3. 3If it looks like a good fit, I send a link to book a free call.

I'll reply within one business day.