A learning lead at a large multinational recently told us how her colleagues get through their mandatory e-learning: playback at 1.5× speed, coffee in hand, quiz at the end, certificate in the system. A dozen topics, a couple of hours a year, every module built by a different owner. Everyone passes.
Then someone in the room asked the question that stuck with us: if their life depended on what they just learned, would they pass?
Silence.
Employees speed-watch compliance training because it restates policies instead of preparing them for real moments. Regulators are moving the same way: NIS2, DORA and the EU AI Act ask for evidence of what people actually do. The fix is a design brief built around rehearsed moments, measured on decisions instead of attendance.
It is tempting to read the 1.5× habit as a discipline problem. It is not. People speed-watch content that was built to be completed rather than used. The playback speed is an honest review, delivered anonymously, by your entire workforce, every year.
Most compliance training is a stack of separate modules, each one owned by a different department, each one restating a policy the reader already agrees with. Nobody disputes that you should not share passwords or accept bribes. Agreement was never the problem.
The problem lives somewhere the modules never go: the four seconds where knowing and doing part ways. The phone call on a Friday afternoon where a client changes a payment account and the voice sounds right. The moment a colleague asks you to sign off on something small, this once, because the deadline is real and the risk feels abstract.
You cannot speed-watch your way through that moment, and you cannot slow it down either. It arrives at full speed, exactly once.
For years the 1.5× habit was survivable, because the thing being audited was the completion rate. That is changing.
NIS2 requires periodic, documented security training. DORA expects procedures to be tested against realistic scenarios rather than acknowledged on paper. The EU AI Act expects people to apply human judgment to AI output, proportional to their role. The pattern across all three: the question is shifting from whether you delivered training to what your people actually do.
A completion rate cannot answer that question. It was never designed to. It proves attendance, and attendance was the one thing nobody doubted.
Here is the counter-evidence to the 1.5× habit. Together with the Dutch Heart Foundation we built a five-minute course on one thing only: hesitation in an emergency. No policy, no quiz, just the moment itself, played out with the learner inside it.
343,000 people completed it voluntarily, and the average session lasted seven minutes. They stayed two minutes longer than the course.
People do not speed past practice that feels like the moment it prepares them for. They lean in, because for once the training is asking them something they do not already know the answer to: what would you actually do?
Fewer restated policies, more rehearsed moments. Episodes that build on each other instead of a dozen silos that each start from zero. Minutes inside the working day instead of hours ripped out of it.
It also needs a sharper definition of a good decision. Almost all training measures whether the answer was right; almost none measures whether it arrived in time. In the moments that matter, right but too slow fails exactly like wrong but too fast.
And the measurement moves inside the experience itself, so you can see how judgment develops from session to session. That record of developing judgment is the evidence the new supervision is asking for.
The 1.5× speed setting is not going away, and the completion rate will keep looking fine. The interesting question is what you give people that they would never dream of speeding up.
&ranj has been building serious games for behavioral change since 1999, with 500+ projects for organizations where conduct carries real consequences. If your mandatory training is being speed-watched, we would enjoy the conversation: get in touch via ranj.com.