A manager I know signed off a full day of customer-service training for her team. Good trainer, decent materials, 100 per cent attendance, strong smile-sheet scores at the end. Three weeks later the calls sounded exactly as they had before. She asked me what went wrong with the course. Nothing went wrong with the course. What went wrong happened after it, in the twenty-nine days nobody had planned for.
This is the quiet scandal of corporate learning. The world is on track to spend around 800 billion dollars a year on training by 2035.1 Most of that money buys a room, a trainer and a feeling of progress, and then the newly learned skill drains away. Hermann Ebbinghaus mapped why more than a century ago: without reinforcement we forget a large share of what we learn within days, not weeks.2 A course is not a change. It is the start of one, and the start is the cheap part.
The transfer gap, in one picture
Estimates of how much training actually shows up as changed behaviour on the job vary a lot, from as low as a tenth to around a third, but every serious review lands on the same verdict: the leak is large, and it happens in the workplace, not the classroom. The reason is simple once you see it. We pour the budget into designing and delivering the course, and almost nothing into the part that decides whether it lands.
Look at that third bar. Transfer, the practice, the manager follow-up, the removal of the old habit that competes with the new one, is where behaviour is won or lost, and it is the line item that never gets funded. You would not launch a product with 95 per cent of the budget on the factory and 5 per cent on getting it to customers. We do exactly that with skills.
What a company that took transfer seriously did instead
Walmart offers the clearest counter-example at scale. Rather than fly supervisors to a hotel for a slide deck, it built around 200 physical Academies attached to real stores and put hundreds of thousands of frontline associates through training on the actual sales floor, with managers coaching the behaviour where the work happens.3 The content was ordinary. The design was not. It closed the gap between learning a thing and doing a thing by refusing to separate them in the first place. That is the whole trick, and it costs discipline more than money.
The solution, as a checklist
Design for the 29 days, not the one
Before you approve any training, make it pass this list. If it fails three or more, you are buying a certificate, not a change.
- The behaviour you want is written as an observable action, not a topic. "Handles an angry customer without escalating" beats "customer service skills".
- Each learner leaves with one specific thing to do differently on Monday, not ten things to remember.
- The line manager knows what to look for and has agreed to look for it in weeks one, two and four.
- The old, competing habit has been named and something has been changed to make it harder to fall back into.
- There is a short reinforcement at day seven and day thirty, not silence.
- Success is measured on the job, by a number that existed before the training, not by a happy-sheet.
Why managers are the real training department
The uncomfortable finding under all of this is that the single biggest predictor of whether training transfers is not the trainer or the platform. It is whether the learner's own manager expects, notices and reinforces the new behaviour. A brilliant course with an indifferent manager loses. An average course with an engaged manager wins. If you have no training department, this is good news, because you already employ the people who matter most. They just have not been told the job is theirs.
Questions to ask before you approve the next course
- What will someone do differently, and how will we see it?
- Who is the manager on the hook for reinforcing it, and do they know?
- What number, already on a dashboard somewhere, should move if this works?
- What are we changing about the day-to-day so the old habit gets harder?
- If nothing changes in thirty days, what will we stop doing?
None of this makes training free, but it makes it honest. Spend the money on the course if you like. Just do not confuse the receipt with the result. The result lives in the month after the room empties, and it belongs to the manager, not the trainer. Fund that month, brief that manager, pick one behaviour, and measure it where the work is done. That is the difference between training you paid for and training that stuck.
Notes and sources
- Corporate Training Market projected to reach approximately 805.6 billion US dollars globally by 2035, at a 7.0 per cent CAGR. Allied Market Research, 2025. Link. ↩
- The Ebbinghaus forgetting curve: newly learned information is lost rapidly without reinforcement. Overview and evidence. Link. ↩
- Walmart built around 200 Academies to train frontline associates on the job. Training magazine, Top 125 best practice, and Walmart corporate. Link. ↩
From the book
Training That SticksA practical method for designing training that moves behaviour, embedding it in the first thirty days, and proving the difference it made, with no training department.