A leader's voice begins with what they notice

Four senior leaders can agree on the same company position and still have no reason to sound alike.

I ran into that problem while developing a thought-leadership programme for one organisation. Each leader needed to speak about related subjects without contradicting the others. But if every article sounded as though it had come from the same corporate desk, publishing under four different names would have achieved very little.

My first instinct was to look for differences in tone.

One leader was more direct. Another reasoned through examples. Some preferred shorter sentences; others were more comfortable qualifying an argument as they developed it.

Those differences mattered. They just weren't the most useful ones.

The stronger distinction appeared in what each person noticed.

Faced with the same broad issue, one leader would move quickly towards the customer consequence. Another would see an operational constraint. Someone else would question an assumption the market seemed to have accepted.

The subject could be shared. Their attention was not.

That changed how I thought about leadership voice.

Tone is the part we can ‘see’

We often talk about voice as though it lives mainly in language.

There is some truth in that. People develop recognisable habits: vocabulary, syntax, sentence length, degree of formality, preference for stories or data. Given enough examples, AI can identify and reproduce many of these patterns.

Research on authorship imitation bears this out. Models and specialised systems can transfer meaningful stylistic features from relatively small samples, although performance varies considerably by type of writing. A large 2025 study covering more than 400 authors found that current models were better at imitating style in relatively structured forms such as email and news than the subtler patterns found in blogs and forums.

So “AI cannot sound like you” is not a particularly useful position. Given enough material and the right constraints, it can get surprisingly close.

What’s more interesting is what it is being asked to sound like you about.

Voice begins with attention

The leaders I've found most distinctive do not necessarily hold provocative opinions on every subject.

They tend to notice particular things.

A founder may repeatedly see the cost of delay. An operating leader notices where an apparently sensible strategy creates work downstream. Someone responsible for customers spots the gap between the language a company uses internally and the way buyers actually describe the problem.

Those habits of attention affect the argument before a sentence is written.

They determine which detail deserves examination. They influence which evidence feels relevant, whose experience enters the story and which consequence is worth spending time on. They also determine what gets left out.

Over enough pieces, a pattern starts to form.

This is one reason I'm wary of beginning leadership-content programmes with a list of topics or “content pillars”. They can be useful organisational devices, but they do not tell us why this person should have something distinctive to say within them.

I would rather understand what reliably catches the leader's attention.

I start interviews with the tension

The least useful opening question in an executive interview is often the most obvious one:

What do you think about this topic?

You usually get an intelligent answer. Maybe even a publishable one. But you rarely get very far beneath the accepted position.

I've found it more useful to begin with the tension. What does the sector get wrong? Which common explanation feels incomplete? Where does what people say publicly differ from what actually happens in practice? That usually gives the conversation some friction.

Then I want the experience behind it. When did the leader encounter this themselves? What happened? What surprised them? Which part of the situation changed how they thought about the issue?

Only then does the opinion become particularly interesting.

The sequence matters because leaders, like everyone else, can produce a polished view before they have worked out why they believe it. A specific experience often complicates the first answer. It adds exceptions, consequences and limits that a generic prompt would never surface.

The interview is therefore doing more than collecting material for a draft.

It is helping locate the argument inside the leader's experience.

Generic thinking can wear familiar language

This is where attempts to “capture someone's voice” often go wrong.

You can analyse ten previous articles, build a detailed style guide and instruct a writer or model to reproduce the leader's sentence patterns. And, honestly, you’ll probably get a plausible piece as a result.

It may also explain the trend, acknowledge the obvious challenges and finish with perfectly sensible recommendations that any competent executive in the sector could endorse. Which is perfectly fine, of course. There’s nothing conspicuously wrong with that.

The problem is that almost any competent executive in the sector could have written it. (And, with AI to help them, most do - LinkedIn is filled with similar posts espousing similar opinions on similar topics.)

The opposite problem appears when distinctive verbal habits are pushed too hard. A phrase the leader uses occasionally appears in every second paragraph. Directness becomes bluntness. A preference for stories turns into a mandatory anecdote at the beginning of every post. You risk making the person a caricature in their own writing.

Distinctive language works when it expresses a recognisable way of thinking. Repetition of the surface alone eventually starts to look like imitation.

There is some emerging evidence for this difficulty. In a 2026 study, 81 participants edited AI-generated drafts for writing tasks where personal style mattered. Editing made the drafts more similar to participants' unaided writing, but the finished text still retained measurable similarities to the original AI output and showed less stylistic diversity than the human-written control material.

For leadership content, that suggests a practical lesson. Trying to recover individuality during the final polish is harder when the underlying argument begins as generic material.

The point of view comes from selection

A good source conversation usually produces too much. There are several relevant experiences. Interesting side arguments appear. The leader remembers a qualification halfway through. Someone sends another piece of evidence after the call.

That’s why a good writer’s job is also to edit.

Which tension deserves the centre of the piece? Which experience helps us understand it? Which evidence supports the claim? Where does the leader actually depart from the obvious answer?

And, perhaps hardest, what are we willing to leave out?

Leadership content often becomes generic because it tries to be comprehensive. Every perspective gets acknowledged. Every adjacent issue receives a sentence. The argument becomes balanced enough that very little is being argued.

Selection gives the piece shape.

It is also where responsibility enters. The leader has to be comfortable with the emphasis, not merely the wording. They need to recognise the claim as something they would still say if a customer, colleague or sceptical peer asked them to explain it.

That is a higher standard than “this sounds like me”.

AI is more useful to me after the conversation

Most of the value I get from AI in executive writing has little to do with imitating style.

After a strong interview, it is useful for organising the material. I can use it to group related observations, test different structures or surface tensions I may have missed in my notes.

I use it to challenge an argument, too. Does the evidence actually support the claim? Is one section repeating another? What would a sceptical reader question? Across a series, it can help identify inconsistencies that become harder to spot when pieces are written weeks apart.

Drafting is part of the process, but it comes relatively late. Easier production moves more value into the decisions around the draft. Leadership writing makes those decisions particularly visible because the named author has to own them.

The model can suggest an interpretation. It cannot settle whether that interpretation is important enough, accurate enough or sufficiently theirs to publish without someone making that judgement.

One company does not need one voice

The multi-leader programme brought this into focus for me.

The organisation needed consistency. Its leaders could not take incompatible positions on matters the company had already decided, or make conflicting claims about its evidence and priorities.

But consistency did not require interchangeability.

A useful system could establish the shared ground: what the organisation believed, which claims it could support and where important boundaries sat. Within that, each leader had different responsibilities, experiences and patterns of attention.

That gave them different routes into the same conversation.

This matters beyond executive vanity. Edelman and LinkedIn's 2025 B2B thought-leadership research, based on nearly 2,000 professionals, found that hidden buyers value thinking that informs or challenges their existing perspective.

A recognisable tone may help someone read the piece. The harder task is giving them a perspective worth associating with the person who wrote it.

The approval conversation should go further

I still ask leaders whether a draft sounds like them.

It catches phrases they would never use. It exposes places where the prose has become too polished, too formal or too far removed from how they actually think.

Before a piece goes out, I want the leader to be able to explain why this issue deserves their attention, which part of it they are asking the reader to notice and what experience shaped the view. They should know where the claim becomes interpretation, and what they would qualify, if challenged.

Most importantly, they should be able to say what they want the reader to reconsider afterwards.

Those conversations tend to improve the language as a consequence. A sentence becomes more specific because the leader remembers an example. A broad claim gets narrowed. A conclusion changes because the original version sounded firmer than the experience justified.

That is where the voice becomes difficult to fake.

The wording matters. But the more durable recognition comes when readers begin to recognise the mind choosing what is worth their attention.