You cannot edit your way to a voice. When a text sounds wrong, the fault rarely lies in the phrasing – it lies in the fact that no one has decided who is speaking. Tone of voice is the sum of a series of choices that precede all writing: What do we hold opinions about, and what do we never comment on? Do we address the reader as an expert, as an equal or as a guide? Which words are ours – and which do we not use, even though everyone else does?

Those choices are strategic in the true sense of the word: they are choices against things. A sender that can talk about anything in any tone does not have a voice but a repertoire. Recognisability does not come from what you are able to say – it comes from what you consistently choose.

Recognisability is a promise

Why is this strategy and not aesthetics? Because the voice is where the reader meets the organisation’s identity in practice. A core narrative that lives in a strategy document but not in the language does not, in reality, exist for the outside world. The voice is the narrative as it sounds on a Tuesday – in an email, a rejection, a job advert.

And recognisability creates obligations. The sender that sounds like itself in its marketing but like an insurance company in its customer service does not have a tone of voice challenge – it has an identity question that shows up in its language. My experience from advisory work is that work on the voice can therefore rarely be confined to the communications department. It ends up being about what the organisation wants, and what it dares to stand by. The language reveals what the strategy leaves unsaid.

The voice in an age of new intelligence

There is now a new reason to take the decision seriously. Language models can today imitate any style on demand – fluently, quickly and endlessly. You might think that made tone of voice less valuable. I would argue the opposite: when everyone can produce smooth, professional text, smooth, professional text is no longer a distinguishing feature. What sets a sender apart is what the machine cannot come up with on its own – the deliberate choices, and choices against, that make a voice belong to this particular sender.

There is a practical side too. An organisation with a defined voice can use the machine as an amplifier: the drafts can be shaped, because there is a standard to shape them against. An organisation without a defined voice gets the average of everything the model has read – I have described the mechanism in more detail in the essay on when the machine writes first. Not because the tool is poor – but because no one has told it who is speaking.

The decision can be trained

Tone of voice becomes strategy the moment it is treated as a decision with an owner: written down, justified and known by everyone who writes – including those who write with a machine. The questions the work begins with are simple to ask and demanding to answer: Could a reader recognise us from three sentences without a logo? Which words do we use that we ought to own – and which do we only use because everyone else does? And who in our organisation has the mandate to say no to a text that is correct but does not sound like us?

How the voice is kept alive when the whole organisation writes with AI is something I have described in the article on when an organisation learns to write with AI – and the work of defining it is part of my programmes on AI for creative work.

Sources

The article draws on the author’s advisory and teaching experience with tone of voice and AI in communication work.