Nobody has time for your training

Industry survey work published for 2026 puts it bluntly: around half of learning leaders and a slight majority of employees say workloads leave no room for training, and 70% of employees admit to multitasking during it. Alongside that sit two figures that explain why nobody escalates it. Nearly a third say their training is too theoretical, and roughly 80% say they tune out when it is not relevant to them. So the picture is not that people refuse to learn. It is that a session has to compete, in real time, against work that has a deadline and a person waiting for it, and a generic session loses that competition every time. Worth noting the provenance: these come from vendor-published survey work rather than academic research, so treat the direction as solid and the decimal places as marketing.

Because it solves for duration and the constraint is attention, not minutes. A fifteen-minute module slotted into a day that has no gaps is not consumed in a gap, it is consumed during something else: on a call with the camera off, while an email draft sits open, between two meetings with the mind still in the first one. The completion rate goes up, which is why the approach keeps getting funded, and the capability does not move, which is why the same problem reappears next year under a new name. There is a narrow case where short works genuinely well, and it is worth being fair about: reference material you consult at the moment of need, which is to say documentation. Looking up how to do the thing while doing the thing is excellent. That is just not training, it is a help article, and calling it learning in the flow of work does not change what it is.

For knowledge recall, less than people assume. For anything requiring practice, fatally. A participant half-listening to a policy explanation will retain some of it, badly, and may well be able to pass the quiz, which is why the metric survives. A participant half-attending a session on handling a difficult conversation learns nothing at all, because the entire mechanism is attempting the thing while somebody watches and corrects you, and you cannot attempt anything while answering Slack. This is the split that should drive every format decision and almost never does. If the session's outcome is that somebody knows a fact, multitasking degrades it. If the outcome is that somebody can do something they could not do before, multitasking does not degrade the session, it cancels it.

The sponsor's, and this is the part that gets avoided because it is uncomfortable. A learning team can control design, relevance and length. It cannot control whether a participant's manager treats the session as protected time or as something to be done around the real work. When a manager says attend if you can, the session is already over, whatever happens in it. So the honest framing of the conversation is not how do we make training more engaging, it is whether the organization is prepared to pay the actual cost, which is the hours, not the vendor invoice. Framed that way it becomes a legible business decision. Framed as an engagement problem, it stays with a team that cannot solve it and generates five years of increasingly clever formats that all fail for the same reason.

Four things that are within reach without new budget. Shrink the catalogue, aggressively, because a smaller number of sessions that genuinely matter is easier to protect than a wide menu nobody can prioritize. Make the sponsor commit the time in writing before the session is built, which is a short conversation that surfaces whether this was ever real. Separate knowledge from practice and stop running the first as a live session, since anything that could be a document should be one and that frees the hours for the things that cannot. And measure the thing you actually want: not attendance, not completion, not satisfaction, but whether somebody can now do the thing, which usually means a short assessment against stated criteria. That last one is also what turns the training record into something defensible rather than decorative.

It is a useful frame and it describes something real. The idea is that organizations have optimised for speed of delivery over depth of mastery for long enough that there is now an accumulated deficit: people have been exposed to a great deal and are fluent in very little. The analogy to technical debt holds in the important way, which is that the interest compounds quietly and the bill arrives as a capability gap at the worst possible moment, usually when something changes and nobody can adapt. Where the analogy breaks is the repayment. You cannot refactor a person. Paying this down means deliberately doing less, more slowly, with practice attached, which is the exact opposite of what a year of pressure to show throughput produces. That tension is the actual strategic problem, and no format solves it.

The honest answer is that the pressure is increasing, not easing. AI adoption has added an expectation that people will learn new tools continuously while output expectations stay flat or rise, and survey work for 2026 reports a meaningful share of employees already worried that leaning on these tools is weakening their own problem-solving. So the demand for learning time is going up at the same moment the supply of it is going down. That is not a cheerful conclusion, but it does clarify the strategy. The organizations that will come out of this well are not the ones with the best content library. They are the ones willing to say no to most of it, protect the hours for the few things that need a room, and accept the trade rather than pretending there is a format that avoids it.