A company can equip an employee’s voice. It cannot own it.

Índice

In late September, Microsoft told its employees that its communications team would leave marketing and report to Brad Smith, whose organisation also handles legal, regulatory and public policy work, according to GeekWire. The memos went beyond reporting lines. Satya Nadella asked for the people closest to the products and customers to become a bigger part of how the company tells its story, and Smith wrote that more employees and executives would likely act as part of Microsoft’s public voice, including on social media. Nadella wanted those voices connected by “a shared strategy and set of principles”. Forbes noted that the memos, as reported, don’t say what those principles are or how much room an engineer will have to post without approval.

That missing detail holds the question I keep coming back to: how much of an employee’s voice can a company shape before it stops being the employee’s voice?

Who the voice belongs to

Forbes shows the other end of the spectrum. Meta’s chief AI officer, Alexandr Wang, posted more than 300 times in the two weeks after the launch of its Muse agent, according to a Business Insider count, with product updates, partnerships and memes in the mix. Forbes itself says most companies would be uncomfortable with that approach. Across the AI industry, product leads and engineers explain and demo their own work to audiences that follow them as closely as they follow the companies they work for. In every one of those cases the company benefits because the voice is personal. A brand can borrow that credibility, and every layer of management applied to it reduces what there is to borrow.

What I would build

If I were leading internal communication in that situation, I would put my energy into material and context before rules. Short videos, visuals and texts that people can use if they want to, a few lines on what we are launching and why, and a short, clear statement of what is confidential. Whether they use the material as it comes, rewrite it in their own words or ignore it completely would be their decision.

Branko Karlezi described a version of this when I interviewed him for this blog. He designed a LinkedIn masterclass for employees that started with the person: “First we invite them to find their tone of voice.” Only afterwards did he show where their interests met the company’s themes. His conclusion was direct: “You can give them tools to communicate, but you cannot, so to speak, moderate or impose communication. That is a bad strategy.” (Both quotes translated from Spanish by the author.) That order interests me, because it puts the kit second.

When a post goes wrong

Picture an engineer who posts about a launch. Two hours later a colleague points out that a figure is wrong, or that the post touches something that was not public yet. It is nine in the evening. She will message her manager, probably on WhatsApp, and the manager will take it to the general manager and to the communications team. I would build the model around that reality. The manager is the first call, so the manager has to be prepared for it: communications trains and empowers leaders to handle the conversation themselves, and to come to us when they cannot answer. That adds a task to a role already carrying plenty, as I wrote in Nobody told managers the job had changed, which is one more reason the preparation has to be real.

Once the manager reaches us, the three of us sit down together: the engineer, her manager and someone from communications. The post stays up. The employee answers, and she writes the answer herself, in her own words. Our part is to look at the reputational risks with her and to supply the correct figure if the one she used was wrong. Nobody looks for someone to blame.

What happens next depends on the reputational impact of the mistake. When it is small, her own answer is enough. When it is larger, the communications team can act from the company’s official account, with her agreement. It can repost her post with the correction, or comment on it to confirm the right information and stand publicly behind her. That comment says what she contributed, for example that she highlighted this or that about the launch, and it can pick up a line from her original post. It does not say that she will not be punished, because announcing that would imply she might have been. The absence of punishment shows in the gesture.

Everyone else in the company is watching how that mistake gets treated. A public response that backs her tells them an imprecise post is not a risk to their careers, and that the company sees live communication as something that makes mistakes and recovers from them. A deleted post, or silence, tells them the opposite. A mistake answered in her own words and backed in public leaves her credibility intact. A deleted post, or an answer drafted by someone else, takes it away.

The test of freedom

The freedom to decline is real only if declining costs nothing. Ahmed Ali El-Shorty offered a test in the comments of my LinkedIn post: if someone ignores what you gave them and nothing happens, you equipped them, and if they get a quiet word afterwards, it was “a script with a nicer name”. Kateryna Byelova added a second condition. People need the freedom to disagree too, because otherwise advocacy starts to look like compliance.

I would add that the quiet word almost always carries good intentions, and it still weighs more than the intention when it comes from a manager or from the communications team, because the role speaks louder than the person. The same happens with what people leave out. When employees can only share good news in approved words, their silence about everything else says more than their posts. The support described above responds to a mistake the employee wants help with. The quiet word responds to a choice the company did not like.

Job titles and job descriptions

Rachel Miller insists that communication never belonged to a single team. In our interview she put it this way: “Internal communication is in our job titles, but it’s in their job descriptions.” If that is true, the engineer who posts about a launch is doing a part of her job, and the function’s task is to make her capable and safe at it.

Miller made a related point about AI that applies here as well. She wants humans at the beginning, middle and end of the process “because we need to put our names to it.” In a post by an employee, the name on it is hers, so the decisions about what to say, how to say it and whether to say it have to be hers too.

Where the line sits

Microsoft’s memos leave the principles unspecified, at least in what has been reported. A company in that position will draw a line between equipping and scripting somewhere, and I suspect it will show in what happens to the employee who ignores the material, writes something different, or says something the company would have preferred not to hear. That detail, more than the kit, tells people whose voice it is.

Voices referenced: Rachel Miller (All Things IC), Branko Karlezi, Ahmed Ali El-Shorty, Kateryna Byelova. Sources: Forbes (Cody Luongo, 6 Oct 2026), GeekWire.

Agregar un comentario

Tu dirección de correo electrónico no será publicada. Los campos requeridos están marcados *

Rellena el siguiente formulario para contactar