Ebbinghaus proved we forget most new learning within days. In today’s overloaded workplace, only immediate practice turns a training insight into lasting change.
We have all sat through training that felt transformative on the day and left almost no trace a fortnight later. That gap between how training feels and what it actually changes has got very little to do with the facilitator. It comes down to memory, and specifically to something psychologists call the forgetting curve.
A 140-year-old curve that still holds
In 1885, the German psychologist Hermann Ebbinghaus ran an experiment on himself. He memorized lists of meaningless syllables, then tested his own recall at intervals ranging from minutes to a month, plotting how quickly the memory faded. The resulting forgetting curve showed retention collapsing fastest in the first day or two, then leveling off into a long, shallow decline. For context, 1885 fell within the second industrial revolution: the motor car was just being invented, and electric lighting was just beginning to spread.
Ebbinghaus’s original data suggests we forget roughly 75% of new material within about a week. A rigorous 2015 replication of Ebbinghaus’s original method, published in PLOS ONE, found that by the six-day mark only around a quarter of the original learning advantage remained, before dropping close to zero by day 31. Different studies land on slightly different numbers depending on the material and the learner, and the exact percentage is less important than the shape underneath it: without reinforcement, most of what training delivers is gone within days, not months.
This is a very uncomfortable thought when so much money goes into training each year, whether the session in question is a two-day leadership offsite or a ninety-minute lunchtime webinar on giving feedback. Most of us already know this instinctively. We just don’t plan or budget for it.
Why 2026 makes it worse, not better
Ebbinghaus ran his experiment alone, in a quiet room, with nothing competing for his attention. So how much do we actually retain in the modern workplace, when we are bombarded with constant notifications and interruptions? Honestly, probably less than most of us would like to admit.
Nobody leaves training and arrives back at their desk to a blank canvas and a silent retreat atmosphere. They arrive back instead to a full inbox, a Slack channel that has kept moving all afternoon, and a calendar that was already double-booked before lunch. Within the hour, three other things have asked for the same fragile, newly-formed memory that training just created, and none of them are the idea the facilitator spent all morning on.
More data and more pressure do not make new learning stickier, whatever the training vendor’s slide deck implies. They make it easier to displace. A training investment that looked entirely sound in the room, with good energy and strong feedback scores, can be functionally gone by the following Monday, not because the content was weak, but because nothing in the following 48 hours gave it a reason to stay.
The one variable that actually changes the curve
Ebbinghaus’s data, and a century of research since, point to a single reliable lever: retrieval. Reviewing notes passively, rereading a slide deck, or watching a recording back barely moves the curve at all. Actively using the material, ideally within a day or two of learning it, does. A recent synthesis in Nature Reviews Psychology confirms that spaced, effortful retrieval produces dramatically better long-term retention than re-reading or cramming, because the act of pulling an idea back out of memory under real conditions is what strengthens it, not the act of putting it in.
You can see this most clearly in immersive training classes built around presentation skills. Almost everyone is reluctant the first time they are asked to stand up and practice communicating in front of a group, in much the same way nobody learns to swim by being told how swimming works. You only learn by getting in the water.
Once the framework has actually been taught, and people are given the chance to practice it three, four, or five times in a single session, with feedback after every attempt, in a room that feels safe enough to get it wrong, the gap between the first attempt and the last is night and day. Nobody leaves that room an expert. They leave having experienced first-hand what a good presentation actually looks and feels like, which is a very different thing from having sat through a slideshow explaining how to give one.
This is the part most training design gets backward. Training delivers a spark: a genuinely useful framework, a moment of real clarity in the room, a feedback model that finally makes sense. What happens in the next 48 hours determines everything, and immediate implementation is the critical point in the whole process.
If that spark is followed by real action within a day or two, actually using the framework on a live, slightly uncomfortable conversation rather than a hypothetical one, it starts to hold. If it is not, the gap fills with false hope instead: the quiet, comfortable belief that because it felt clear on the day, it will still be there when it is needed three weeks from now, in the room, under pressure, with no notes. It will not. Put simply: if you do not use a new skill fast, you lose it, and quicker than most people think.
What this means for anyone commissioning training
The practical implication for leaders isn’t to run better training classes, or to hire a better facilitator. It’s to stop treating the training as the whole intervention. The long-standing 70-20-10 model for workplace learning breaks development into three sources: roughly 70% from hands-on experience and practice, 20% from feedback and coaching by managers and peers, and only 10% from formal training such as a course. A single session was never designed to carry the weight most training budgets put on it, because it can only ever supply that final 10%.
Three changes make more difference to what survives than almost anything spent on the day itself.
- First, build a plan that structures the practice: a specific task on real work, within 48 hours of being in the training room, not a vague intention to use it when it comes up.
- Second, build in a short check-in a week or two later to get feedback on where the new skill is going well and where it needs to pivot, rather than assuming the training either landed or it did not.
- Third, practice the skill again in light of that feedback, so it is refined rather than abandoned. That sequence: plan, practice, get feedback, pivot, practice again, is what the 70-20-10 split is actually describing, and it costs far less than the training that triggered it. None of it requires a bigger training budget. It just requires someone to plan for the week after the training, which almost nobody currently does.
This doesn’t mean training classes are a waste of time or money. It means they cannot be the finish line, and most training budgets currently treat them as exactly that. The value of a training day is decided in the days that follow it, by whether anyone used the idea before the forgetting curve had its say.
By Jeremy Campbell