In the early 1970s the US Army braced for a catastrophe that never arrived, because doctors in Vietnam were reporting that a considerable share of soldiers were addicted to heroin, and the fear was that the returning men would carry a wave of addiction home that no therapy of the day could stop. The psychiatrist Lee Robins followed the men after their return and found a number that contradicted the addiction research of her time: around twenty percent of the soldiers said they had felt addicted in Vietnam, yet of those addicted men only five percent were addicted again in their first year at home, and even over three years only twelve percent. When Robins looked back on the study in 1993, she titled her lecture with a question, whether it had been "a fluke or normal expectation".
The most effective withdrawal anyone had observed up to then consisted of a flight home, and the remarkable part is less the drug than the direction, because a behavior that seemed unbreakable in one place dissolved as soon as the place changed and the environment at home no longer supplied any of the cues that had carried it over there. Learning in a company almost never means knowledge alone, it means that somebody does something differently on Monday than on Friday, and every training programme passes through the same mechanism in reverse on its way there: the new behavior is produced in a place built specifically for it, the seminar room, the learning platform or the serious game, and then the participants fly back into an environment whose cues, incentives and metrics have carried the old behavior for years.
The missing behavior is already a finding
Every company that buys a training programme has spotted a gap first: managers give feedback too rarely, sales sells discounts instead of solutions, teams do not record how they reached a decision. The usual diagnosis says people lack something, knowledge, technique or attitude, and the training is supposed to supply it. That diagnosis skips a question more uncomfortable than any needs analysis, namely why the desired behavior has been missing so far even though many of those involved have long been capable of it, because a behavior that is missing in an organization is rarely missing by accident, it is missing because the environment never demanded it, never made it easier and never rewarded it.
Robert Mager and Peter Pipe put a test question to this as early as 1970, one that almost everyone in training knows and hardly anyone follows to its end: could the person do it if their life depended on it? If the answer is yes, the problem is not a training problem, and every further seminar treats a symptom in the wrong place. A no does make training necessary, but it only does the first half of the work, because the freshly learned behavior then returns to exactly the environment that has explained its absence so far.
A behavior that is missing in an organization is rarely a hole in its people's knowledge and almost always an imprint of the environment they work in.
The industry's most famous number was never measured
Anyone who has worked in training for a while has heard often enough that only ten percent of what is learned reaches everyday work, and the sentence sounds so familiar that people take it for a research finding. It goes back to a 1982 article by David Georgenson, who opened with the rhetorical question of how often his readers had heard training directors say that only about ten percent of the content presented in the classroom later showed up in behavior, and Georgenson cited neither data nor source, because an opening line did not need any. Kevin Ford, Stephen Yelon and Abigail Billington traced in 2011 how that opening line turned into a doctrine and titled their paper accordingly with "The 10% delusion".
The more robust number is hardly more comforting: in 2006 Alan Saks and Monica Belcourt asked training professionals from 150 organizations to estimate how many employees apply what they learned, and the answers came to 62 percent immediately after training, 44 percent after six months and 34 percent after one year. These too are estimates rather than measurements, and the fact that an industry has to answer its most important question with an estimate and an invented number says more about its measurement habits than about its participants. Apply the estimate to the figures of the German Economic Institute (IW), according to which companies in Germany invested around 46.4 billion euros in training in 2022, 1,347 euros per employee on average, and a company with 400 employees arithmetically leaves around 355,000 of its roughly 540,000 euros in annual budget behind in the seminar room.
Where what people learn is outvoted
The new behavior does not vanish on Monday, it is outvoted, and Engaginglab calls the point where this happens the transfer break. It sits almost always in one of three places, first in the default, because in the ticket system, the template and the approval process the old way is the preset one, so anyone who wants the new way has to decide against the preset every single time until at some point they stop. The second place is the incentive, because target agreements, bonuses and promotions keep rewarding the old behavior, and the third is visibility, because nobody sees whether someone does it differently, and what nobody sees does not count as normal. In two diary studies, Wendy Wood and her colleagues classified 35 to 43 percent of everyday actions, depending on the study, as habits, as behavior performed almost daily in the same context, and that context is exactly where the seminar sends its behavior back to.
The market has known the problem for decades and answers it with transfer management, with reminder emails, learning buddies, transfer coaching and follow-up modules that all start at the same point, the person, and that all work for as long as someone keeps reminding. Transfer infrastructure starts at the environment, at defaults in tools and processes, at incentives, approvals and metrics, at roles and visible norms, and it works precisely when nobody thinks about it anymore. Geary Rummler and Alan Brache put the reason into a sentence that could hang above the door of every training department: "If you pit a good performer against a bad system, the system will win almost every time."
The strongest objection to this view comes from the research itself, and it deserves an honest answer. In 2010 Brian Blume, Kevin Ford, Timothy Baldwin and Jason Huang analysed in a meta-analysis of 89 studies what predicts transfer, and the person is anything but a side issue there, because cognitive ability correlates with transfer at around .37, somewhat more strongly than supervisor support at around .31, which is why the authors explicitly find "no clear superiority of individual variables over situational variables, or the reverse". That is exactly where the argument lies, though, because training has always worked on the person, through selection, didactics and motivation, while the environment, which statistically explains roughly as much, belongs to nobody in most organizations.
Transfer management asks how to remind participants of what they learned after the training, and so it works for as long as somebody reminds them. Transfer infrastructure asks what in the environment has carried the old behavior so far, and rebuilds defaults, incentives and visibility so that the new behavior becomes the easier path.
The first question can be answered in the seminar room, the second only with IT, controlling and leadership at the same table.
One sales training, built twice
A typical case, assembled from patterns that turn up in sales organizations again and again, shows what this looks like in detail: a company sends sixty salespeople to a two-day training on consultative selling, where they learn to clarify needs before writing a quote, to aim for fewer deals of higher value and to use the discount as the last resort rather than the first, and by the end of the two days they can do it, as the role plays impressively show. On Monday they open the CRM, in which the quote template is the first button after a contact is created and the needs-analysis field is optional, commission is still paid monthly per closed deal, and the weekly meeting displays the ranking of calls and closed deals, on which a thorough needs analysis only shows up as a gap. Three weeks later the team sells the way it did before, and the training evaluation still shows excellent satisfaction scores, because the two days really were good.
The same training, embedded in a transfer infrastructure, looks different mainly outside the seminar room, because the CRM only unlocks the quote template once the needs analysis has been filled in, commission gains a component for deal value or follow-on business alongside the close, and the weekly meeting shows how many needs analyses were held and which order of the week came out of them. None of these changes is didactic, most of them cost a few days of configuration and a tough negotiation with sales controlling, and yet each of them decides more about the success of the training than the two seminar days themselves.
The dashboard of the training department
The training department is itself steered by a dashboard, and almost everywhere this dashboard measures the seminar room: participant numbers, learning hours per head, completion rates on the platform and satisfaction at the end of day two. Like every metric, this one educates, and it educates L&D into a buyer who procures programmes, fills them and gets them rated well, while Monday, on which the value of the programme is decided, appears in no row at all, a pattern we described for metric systems in general in the article on the defensive dashboard. Robert Brinkerhoff, who developed one of the industry's most pragmatic evaluation methods with the Success Case Method, wrote the decisive sentence in 2005: "it is performance, not HRD, that achieves (or does not achieve) results", which is why, he argued, any impact evaluation has to inquire into the performance context and not only into the training.
Who builds Monday
Training is the only function in a company that knows both rooms, the seminar room from the inside and Monday from what its participants report back, which makes it the obvious builder of the way back. For L&D the work shifts in three places as a result, first in the needs analysis, which asks not only what people lack but also what the environment has prevented so far, then in the budget, in which a fixed share of every programme is reserved for adjusting tools, target agreements and metrics, and finally in the role itself, which sits at the table with IT, controlling and leadership before the first module is booked. Only at this point does a learning programme become efficient at all, because the money no longer has to work against the environment in which it is supposed to take effect.
Lee Robins' soldiers left their behavior behind in Vietnam because the flight home took them into an environment that did not carry it, and every training programme puts its participants on the same flight home, only with the opposite wish. What remains open is who built the place where they land on Monday.