Learning is not the same as knowing

A training day on commercial law, ten participants from procurement and sales. Work out the cost: 80 person-hours, plus the fee. Four weeks later the first defective delivery arrives. Nobody examines it, nobody gives notice of the defect. In a commercial sale the goods are then deemed approved for every defect an examination would have revealed, under Section 377 of the German Commercial Code (Handelsgesetzbuch, HGB).

The training day was not the problem. The duty to give notice of defects was covered at length, with deadlines and examples, and by the end of the day everyone could explain it.

Knowledge is not something you acquire once and then own. It is something you repeat until it is simply there. Anyone who overlooks that when planning a training buys a day on which everyone learned something, and a quarter in which nobody knows it anymore. The training is the beginning, and anyone who plans it as the end wastes it.

Forgetting is the normal case

Hermann Ebbinghaus measured on himself how fast what has been learned disappears, and the result is still called the forgetting curve.

How fast what was learned disappears

The largest part is lost on the first day; after that the loss slows down, but it does not stop. Ebbinghaus measured this by the time he needed to learn the same material a second time. In 2015 his experiment was repeated, and the beginning of the curve had the same shape again.¹

Two things belong with the curve so that it does not claim more than it can. Ebbinghaus learned nonsense syllables, and the repetition of his experiment ran with a single subject as well. A rule with a case behind it is something other than a string of syllables, and about that these experiments say nothing. The direction stays the same all the same, and anyone who knows it stops being surprised by the four weeks from the beginning.

Repetition at intervals is what stays

The means against it is unspectacular. Retrieve the same material again at intervals, then again, then again. Anyone who repeats like that keeps considerably more than someone who spends the same time on a single day; that is well documented.²

Anyone who goes through the duty to give notice of defects three times in half a year on a case of their own knows it. Anyone who heard it once does not.

Less of it survives daily business

So far this sounds like a rule that works by itself. It does not. In studies where the material was not practiced on its own but inside a running course next to everything else, the effect came out smaller. The difference is not statistically established, but it points in the same direction in both investigations.³ Why that is, the authors themselves do not know for certain. One of their explanations is that anyone who has the materials at hand in a course looks things up instead of retrieving them, and looking up does not train remembering.

For planning I draw two conclusions from that which go beyond the studies. The repetition has to be a retrieval, a question with no slide deck next to it. And it has to be scheduled, because left alone, the day-to-day pushes out every repetition. That is exactly why it belongs in the planning of a training, not in the good intention afterward.

A limit applies here as well. The studies come from classrooms, lecture halls, and laboratories, not from commercial law, and what is proven is the interval, not the precedence of the question over rereading, at least not in the study on mathematics learning I draw on here.⁴ What carries over is the direction, and that is enough for a planning decision.

I ask three questions before every training proposal

First: what should people do differently after the training? Not know, do. If there is no answer to that, what is needed is a decision, not a training.

Second: when does the first real application come up, and what holds the knowledge until then? The case rarely arrives the day after the training. Even four weeks are too many if nothing happens in between.

Third: who repeats with whom, at what interval, on which documents? Without that answer, the series is merely an intention.

What I ruled out is not the training day but the training day without a follow-up, because it ends the effect at the point where it first arises. How I can tell whether it has worked is not the feedback at the end of the day. It is the call six weeks later in which someone flags a clause they would not have noticed before.

The strongest objection comes from consulting

The hardest criticism of my proposal comes not from learning research but from David Maister, an authority on the management of law firms and professional service firms. Maister was paid to deliver training for years, with high ratings and repeat bookings, and still considers most business training a waste of money and time.⁵

His diagnosis goes further than mine. For Maister, training is an excellent last step in changing behavior and a useless first one. If management has not settled beforehand which behavior it requires, what it measures, and what it rewards, then even the best series changes nothing. And the best training, he writes, is usually done by the firm’s own practitioners; outsiders are there only to develop programs and to train the trainers.

I disagree, but only halfway. On managerial behavior he is right, and on subject knowledge that already exists inside the company he is right as well. Commercial law is a different case, because the knowledge is precisely what is missing internally. Someone who does not know the duty to give notice of defects cannot teach it to anyone either. Why that gap rarely surfaces is set out in the article “Commercial law as a knowledge gap in companies”. What remains of Maister’s objection is the manager in the room, as a participant, not as a guest. That costs one entry in a calendar and changes how seriously the matter is taken.

How the knowledge stays in your team

From repetition at intervals follows what you may demand of a training if something of it is to stay. That means three things for what I offer.

After a webinar the training does not end with the session. The participants can ask follow-up questions for 14 days, and the questions that come then arise from their own contract instead of the slide example. That is the opportunity for the second repetition.

For an in-house training I recommend two days instead of one. That leaves time for questions and for exercises on your own cases, and whether one day or two is your decision. On top of that I offer two refreshers: online sessions of one hour each, the first after six weeks, the second in the sixth month afterward, on new cases and on the questions that have come up in your team by then. The same topics, a second and a third time, each time once the first cases of your own have run.

And for what nobody needs to keep in their head but must be able to find, there is knowledge management. Knowledge that can be retrieved where decisions are made does not need to be repeated. Why that is the infrastructure question for every company that works with contracts is set out in the article “AI gives answers. Your company needs the right ones”.

What sets these formats apart from other offers is less the repetition than the material that gets repeated. I deliver the training myself, as a specialist lawyer for international business law who negotiates and reviews contracts like yours every day, with cases from my own practice instead of a standard slide deck. I am one of two ICC-registered Incoterms 2020 trainers in Germany. What your team repeats are therefore not textbook examples but the cases it may have itself next week.

Back to the 80 person-hours from the beginning. They are not the problem, they are the investment. The question is what stands next to them. Two refreshers of one hour each cost the same ten participants 20 person-hours, plus the fee. That is a quarter of the training day, and that quarter decides whether the day achieves its purpose. There is no cheaper decision in planning a training.

Notes

1. Murre/Dros, PLoS ONE 10 (2015), e0120644. What is measured is the savings on relearning; the percentages in the text and the chart are the remainder to 100. For Ebbinghaus the savings are 58 percent after 20 minutes, 34 after one day, and 21 after 31 days; in the replication 47 and 32 after 20 minutes and one day. ↩

2. Latimier/Peyre/Ramus, Educational Psychology Review 33 (2021), 959. Spaced against massed retrieval across eleven studies: g = 0.74 after correction for publication bias; expanding against uniform intervals across 16 studies: g = 0.034, with no established difference. ↩

3. Murray/Horner/Göbel, Educational Psychology Review 37 (2025), Article 75, a meta-analysis on mathematics learning. Material practiced on its own g = 0.43, material inside a running course g = 0.24, from two groups of different studies; the authors call the distance numerically, not significantly, smaller. The same direction in Bego et al., International Journal of STEM Education 11 (2024), Article 9: a significant effect in two of nine introductory courses; the pooled analysis is significant, but no longer once the mathematics course is removed. ↩

4. Retrieval against plain rereading showed no established difference in Murray/Horner/Göbel (note 3): g = 0.18, with a confidence interval crossing zero. For other subjects the authors themselves point to the well-documented effect of retrieval. ↩

5. David Maister, Why (Most) Training is Useless, 2006. ↩

Reference: Poleacov, P. (2026). Learning is not the same as knowing. INN.LAW. https://inn.law/en/perspectives/in-house-training-forgetting-curve/

---