"But I'm the chairman! They'll pay attention to me!"
He said it with total conviction. We had just advised him to keep his all-hands video message to six minutes or less — or to break it into smaller pieces — because we had viewing analytics from tens of thousands of corporate videos telling us what was about to happen. He waved the advice off. He was the chairman. The rules of attention did not apply to him.
So we shot the clip he wanted and posted it. Eight thousand employees clicked play.
At the six-minute mark, 4,200 were still watching. By our measure that was a normal result. But by the end, 1260 were left — about sixteen percent of the people who started. His audience behaved precisely the way the data said they would, and when we showed the numbers to the Director of Corporate Communications, he went pale and said, in effect, that the data should disappear before it cost him his job.
The data didn't disappear. The chairman was gone within the year. But the lesson outlasted both of them, because it was never really about that one man's ego. It was about a curve that every corporate video obeys — and a curve that has been getting steeper ever since.
Attention follows a law
Across the tens of thousands of corporate videos we have measured, the pattern has barely varied: by the twelve-minute mark, only about sixteen percent of the original audience is still paying attention. Not sixteen of a weak video. Sixteen percent of the average one — polished, well-lit, professionally produced, delivered by someone the organization has every reason to listen to.
The chairman believed authority would hold attention. It doesn't. Attention isn't a function of who is speaking or how important the message is. It's a function of format. A long, linear, passive video makes one enormous assumption about the person on the other end: that they will sit still and absorb, minute after minute, with nothing asked of them. Fewer and fewer people will. The seniority of the speaker has no bearing on it.
This isn't a REACHUM theory. It's one of the most replicated findings in attention science, with a name: the vigilance decrement — the well-documented tendency for sustained attention to erode the longer a task runs, a decline measurable within the first few minutes when the material is demanding. And it has been observed in the exact situation that matters here. When researchers filmed people watching a forty-minute lecture video, both their attention and their later memory of the material fell steadily as the minutes passed, while mind-wandering climbed. The audience didn't decide to stop caring. Their attention drained on a schedule — the way it always does.
The numbers haven't held steady. They've gotten worse.
That chairman's video is years behind us now, and we never stopped measuring. The curve hasn't flattened — it has steepened, and we can put a number on it. Over the last twelve years, the median persistence on a video — how long the typical viewer lasts before dropping off — has fallen from six minutes to two minutes and thirty-six seconds. That's nearly sixty percent of a typical viewer's staying power, gone in just over a decade.
Sit with that against the chairman's demand. He insisted on twelve minutes of attention. Today the median viewer is gone at 2:36 — less than a quarter of what he asked for, and well before the part of any message that actually matters. The cliff that humbled him doesn't just arrive sooner now; for half the room it arrives before the speaker has finished warming up.
We're not the only ones watching the floor drop. The video-analytics firm Wistia, whose State of Video benchmarks draw on tens of millions of videos, recently logged its steepest single-year engagement decline in four years — overall engagement down roughly seven percent, and short three-to-five-minute videos down ten. Our 2:36 figure is sharper than any industry average because it's drawn from corporate training specifically, and from a single consistent measurement method over twelve years. But it points the same direction the broader data does: the attention you could once assume is thinning, measurably, year over year.
This is the part most organizations miss. They treat the completion problem as if it were stable — a known tax they've always paid and can keep paying. It isn't stable. It's compounding. Every quarter you keep shipping long linear video, a larger share of your workforce checks out earlier, and the gap between "we delivered the training" and "they actually absorbed it" widens.
Why "make a better video" is the wrong fix
The instinctive response to a collapsing completion curve is to spend your way up it: higher production value, a sharper script, a more charismatic presenter, motion graphics. We have watched companies pour real budget into exactly this, and the curve barely moves — because production quality was never the broken variable.
The broken variable is the unit itself. A twenty-minute video asks for twenty minutes of uninterrupted passive attention from people who no longer have it to give. Polishing that video is optimizing the wrong thing. You can make the most beautiful object on earth and it will still shed two-thirds of its audience before the halfway point, because the audience isn't rejecting the quality. They're rejecting the demand.
The McKinsey-style instinct — study it, restructure it, produce it more expensively — runs into the same wall the chairman did. You cannot out-produce a format that's fighting human attention. You have to stop fighting it.
Build for the attention people actually have
Attention doesn't vanish at the 2:36 mark. It resets. The moment you ask someone to do something — answer, choose, respond, retrieve — you pull them back from the edge of the cliff and start the curve over. That single insight is the difference between content people endure and content people complete.
And this, too, is in the research — arguably the most useful finding of all. In one sustained-attention experiment, the vigilance decrement disappeared entirely when passive monitoring was replaced with an active task: ask people to click, decide, and respond, and the decline that had looked inevitable simply didn't happen. Classroom eye-tracking shows the same effect — genuine interaction held attention above ninety percent on-task for their full length, not just the opening ten minutes. Interaction doesn't decorate learning. It's the mechanism that keeps attention alive.
It's why the durable answer to the completion problem isn't better video. It's a different shape of learning altogether: short, interactive, chunked into segments that respect how attention actually behaves, with retrieval and response built into the flow rather than tacked onto the end — ideally inside the tools and chat surfaces people already use, not in one more platform they have to be dragged into. Microlearning, done properly, isn't "video, but shorter." It's learning designed around the attention curve instead of in denial of it.
That premise is the reason REACHUM exists. The platform is built attention-first: interaction that resets engagement before it decays, content delivered in segments people finish rather than abandon, and completion and comprehension measured in real time — so you see the curve as it happens instead of discovering, like that Director of Communications, that the numbers you were afraid to look at were true all along. When the format works with attention, the fifteen-percent completion rate that humbled a chairman stops being a law of nature and starts being a solved problem.
The data doesn't care about your title
The chairman's mistake wasn't vanity, exactly. It was assuming the rules bent for him. They never do. The viewing data is indifferent to seniority, to budget, to how important the message is or how much everyone agrees it should be watched. It only reflects what people actually did.
That indifference is a gift, if you let it be. It means the path to training that lands isn't more authority or more production money — it's humility in front of the curve, and a format honest enough to be built around it. Design for the attention your people actually have, and they'll stay. Insist they pay attention because you outrank them, and the data will quietly, ruthlessly prove you wrong.
See what attention-first learning looks like at reachum.com.