Most training is judged on the day itself. The real test is what people are doing differently three weeks later.

In 1885, a German psychologist called Hermann Ebbinghaus published the results of an experiment on the one participant he could guarantee would turn up: himself. He invented a pool of around 2,300 meaningless syllables—WID, ZOF and their friends—memorised lists of them, and then tested himself at intervals to see what had survived.

What he found became known as the forgetting curve. Loss is steepest straight after learning, then flattens out. On his figures, roughly a third of the original effort remained after a day. There are caveats, of course: one man, nonsense syllables, and meaningful material fades more slowly. But the shape has been replicated many times since, including a careful rerun in 2015 that produced a closely matching curve.

It describes, with uncomfortable accuracy, what happens to a great many training days in financial services. You know the scene. Delegates leave buzzing. On Monday they face 200 unread emails, a client call and an IT system that has decided to misbehave. By Friday the photo of the flipchart is in a folder called “Training”, where good intentions go to die.

We’re measuring the wrong thing

Ask most firms how they evaluate training and you’ll hear about attendance, completion rates and feedback forms. These tell you the room was comfortable and the biscuits were adequate. If the training is online and delegates were working from home, neither may be the case. They don’t tell you whether anybody did anything differently.

An attendance record shows that a session took place. It is not evidence that competence improved.

I’ve reviewed various studies on training effectiveness, including suggestions that only around 10% works. Honestly, this figure is shaky and varies by industry sector and research methodology. The World Economic Forum’s Future of Jobs Report 2025 surveyed how employers plan to assess skills through 2030. The hierarchy:

  • 81% will rely on work experience—the most common method, and rising in importance.
  • 48% will use skills assessments.
  • 43% will continue to use a university degree as a requirement.
  • 17% will prioritise apprenticeships.
  • 14% will consider online certificates in hiring decisions.

But the pattern is similar.

That is a better picture, but look at its shape. It’s Ebbinghaus with a lanyard. Whatever the true number is, and your organisation may well differ, learning fades unless something holds it in place: that is application. The evidence base is consistent:

  • Role-play and simulation: 34% better skill application than lecture.
  • Spaced reinforcement: 60% better knowledge retention.
  • Microlearning: 80% improvement in long-term retention.
  • Peer-to-peer learning: 55% increase in knowledge retention.
  • Regular coaching cadence: 50% higher likelihood of exceeding quota.
  • Applying skills within 24 hours of learning: 70% higher retention.

This matters beyond the training budget. The FCA’s guidance in the Training and Competence sourcebook expects firms to assess training needs at regular intervals, to review the quality and effectiveness of the training they provide, and to review employees’ competence regularly and frequently. The wider competent employees rule in SYSC applies well beyond retail activities. An attendance record shows that a session took place. It is not evidence that competence improved.

The event isn’t the problem

Here is the twist: the workshop is usually the best-designed part of the process. The trainer—including this one—is skilled, content is sound and exercises are relevant. What’s missing are the bits either side.

It’s the gym membership problem. A superb personal training session in January, followed by no routine, no accountability and no next session, will produce a lovely afternoon and little else. People don’t blame the personal trainer for the abandoned membership in March.

So if the event isn’t the problem, what would a training programme that survives Monday look like? Here are four things you can do.

1. Brief the manager before, not after

Baldwin and Ford’s model of transfer puts the work environment alongside the individual and the training design, and later research has kept pointing to manager support as one of the strongest influences on whether learning gets used. Yet managers are often the last to know what their team has been taught.

Give them one page: what the course covers, the two or three behaviours to look out for, and one question to ask afterwards. Ask each delegate to arrive with a real situation they want to handle better. It costs almost nothing.

2. Make it real

Generic case studies about a fictional company are a splendid way to learn about that fictional company. If you want a difficult client conversation to go better, practise the difficult client conversation: explaining a fall in a fund to an anxious client, or the compliance officer challenging a senior colleague’s shortcut. Build scenarios from your own firm’s situations, suitably anonymised, with awkward pauses left in.

3. Space it out

Ebbinghaus didn’t only identify the problem. His work also showed that review is the cure, and that each well-timed revisit slows the next round of forgetting. So build the revisiting in: a short prompt after two days, another after two weeks and another after six, a 30-day action commitment, or a practice partner from the same session.

A PDF summary is not a revisit, but the sort of thing people mean to read.

4. Ask on Monday

This is the cheapest and most powerful step. Within a week, the manager holds a five-minute conversation: What have you tried? What happened? What will you try next? No form and no system. A conversation that says, in effect, “this mattered enough for me to ask”.

Measure what you actually want

If the aim is different behaviour, measure behaviour. Sample calls or meetings against two or three specific behaviours. Using AI meeting summaries and recordings makes this process way easier than the meetings I used to observe when I worked in banking. Ask managers what they’ve noticed. Have delegates rate themselves before and after, then compare those ratings with the manager’s view. The gaps are often the most interesting bit.

Evidence like this is far more persuasive than a completion certificate when someone asks how you know your people are competent.

Designing the Monday

The forgetting curve is well over a century old. The cure—review, practice and support—is barely younger, and needs no new software licence. What it requires is deciding that the training event is the start of the process rather than the whole of it.

A simple test: before your next programme, write down what should be different by Monday week and who will notice. If you can’t, you haven’t designed a training programme. You’ve booked a day out.

References

  • The forgetting curve: Hermann Ebbinghaus (1885)
  • Baldwin & Ford (1988), Transfer of Training: A Review and Directions for Future Research, Personnel Psychology
  • Georgenson and the 10% figure (Work-Learning Research)
  • World Economic Forum: The Future of Jobs Report 2025
  • FCA Handbook, TC 2.1 Assessing and maintaining competence
  • FCA: Training and competence