Everyone lists communication. Nobody teaches it.

An analysis published in February 2026 looked at more than 200 employer-authored job postings with AI in the title, drawn from a major job board, spanning technical, research, healthcare, marketing and operational roles with no filtering by industry, location or seniority. It found that 83% of those postings named at least three of five soft skills: strategic and conceptual thinking, critical thinking, communication, attention to detail, and creative problem-solving. The detail worth pausing on is how communication was characterised, which was coordinating action with both humans and AI systems, alongside expressing ideas clearly and interpreting inputs. Two caveats before anybody builds a strategy on it: 200 postings is a modest sample, and the analysis was produced by a company selling hiring assessment tools, which is a party with an interest in soft skills being measurable and important.

Because it is not the thing communication training teaches. The ordinary corporate course under this heading covers presenting with confidence, writing clearly, active listening and adapting your style to your audience. All reasonable, all about the transmission of a message from one person to another. Coordinating action is a different problem. It is working out who needs to know what, in what sequence, with what level of detail, so that a group of people and now a set of systems move in the same direction without a meeting about every decision. That is logistics and judgement rather than eloquence, it is mostly invisible when done well, and the people who are good at it are rarely the most articulate people in the organization. They are the ones whose projects do not stall.

The behaviors that make coordination work, which are concrete and trainable. Framing a request so it contains the decision, the deadline and the consequence of missing it, rather than context followed by hope. Asking the second question, because the first answer is nearly always a summary and the thing you needed was underneath it. Writing the update that the reader can act on without replying to ask what you meant. Noticing when agreement in a meeting is real and when it is social, which is a distinguishable thing if you know what to look for. Closing a loop you opened. None of these appear in a communication skills curriculum and all of them are what people mean when they say somebody is good at communicating. They are also, almost exactly, the moves a facilitator makes in a room.

It raises the cost of being imprecise, which is a genuine shift rather than a slogan. A colleague who receives a vague instruction will usually fill the gap sensibly, because they have context you did not supply and an interest in the outcome. A system will fill the gap too, but with a plausible guess, and it will do it at speed and at volume. So the penalty for ambiguous framing has gone from a minor delay to a confidently wrong output that somebody has to find and unpick. That is why employers are describing communication as coordinating action with humans and systems rather than as presenting well. The irony worth naming: the skill people assumed would be protected from automation is being reshaped by it, and the reshaping is making it more demanding rather than less.

Because it is unfalsifiable as usually framed. Communication skills is a category, not a behavior, and you cannot assess a category. So a course gets built around models and self-awareness, participants rate it well, everybody records a development objective, and nobody can tell afterwards whether anything changed, because nothing specific enough to change was named. The fix is unglamorous: pick behaviors narrow enough to observe, practice them where somebody can see, and assess against criteria that a second person would apply the same way. That is more work than running a workshop on communication styles, and it produces a record that holds up when somebody asks what the training actually did, which is increasingly the question being asked.

Substantially, yes, and I am aware that is a convenient conclusion for somebody who teaches facilitation, so take it with that in mind. But the overlap is not rhetorical. The facilitator's core moves are structuring who speaks when, surfacing what is not being said, making a decision explicit rather than assumed, and closing the loop so the room leaves knowing what happens next. Strip the room out and those are the same moves that make distributed, asynchronous, partly automated work function. The difference is only that a facilitator does it live and visibly while a well-coordinated team does it continuously and invisibly. If employers are hiring for the second thing and nobody is training it, the people who learned it in the first setting have a transferable advantage that is not widely priced yet.

If you are building a development plan, stop writing communication as an objective and write a behavior instead, because the objective as written cannot fail and therefore cannot succeed. If you are hiring, stop testing for articulacy in interviews, which selects for people who perform well in a forty-minute conversation, and give candidates something to coordinate. If you are running learning and development, look at what your communication offering actually trains and ask whether a participant could do anything differently on Monday as a result. And if you are the person who keeps getting told you are a strong communicator without knowing why, it is probably not your speaking. It is that people know what to do after they talk to you, which is worth knowing because it is the part you can deliberately get better at.