Somewhere in your L&D dashboard is a number that feels like good news. A feedback score. A completion rate. Maybe both are excellent — 4.8 out of 5, 94% completion, glowing comments in the exit survey. By every measure your team tracks, the programme worked.
Then Monday arrives, and nothing on the floor looks different.
This isn't a contradiction. It's what happens when the metric and the outcome were never actually connected in the first place.
Completion measures attendance. It doesn't measure capability.
A learner can sit through a full session, answer every quiz question correctly, rate the facilitator highly, and still be unable to handle the exact situation the training was meant to prepare them for. That's not a design failure you'd catch by looking harder at satisfaction scores — because satisfaction scores were never measuring the thing that matters. They measure whether people enjoyed the experience of being told something. They say almost nothing about whether people can now do something they couldn't do before.
The research bears this out more starkly than most L&D conversations acknowledge. Published estimates of training transfer — how much of what's taught actually shows up as changed behaviour on the job — typically fall between 5% and 20%. One well-cited study found that behaviour change measured shortly after a training programme fell from 62% to 34% within a year. The number that people feel good about walks straight back down within months.
Where does the other 80% go?
Not at the point of spending. The money isn't lost when the training is delivered — it's lost in the weeks after, when nothing gets practised. A workshop can be excellent and still fail to build capability, for a simple reason: knowing what good looks like and being able to produce it under pressure are two different skills, built two different ways.
Think about the last time you actually got better at something difficult — a negotiation, a hard conversation with a direct report, a technical judgment call. It's unlikely you got better by watching someone else do it well. You got better by doing it yourself, badly at first, with something at stake, and adjusting. That loop — attempt, consequence, feedback, repeat — is how capability is actually built. It's also the part almost entirely missing from a well-produced video module or a well-facilitated workshop that ends when the room empties.
A different question for your next programme review
Most L&D reviews ask: did people finish it, and did they like it? Those are legitimate questions. They're just not the ones that predict whether behaviour changes.
The more useful question is narrower and less comfortable: if I picked five people at random from this cohort, right now, and put them in the exact situation the training addressed, would they respond differently than they would have before? Not "could they describe the right approach" — would they actually do it, under the pressure of a live moment, with someone watching?
For most programmes, the honest answer is: we don't actually know because we never built anything into the programme to test it.
That's not a criticism of the people designing the training. It's a structural feature of how most corporate learning is built — content-first, consumption-first, with practice treated as optional reinforcement rather than the point of the exercise.
What would it take to design it the other way round — starting from the real decision someone has to make on the floor, and building everything else backwards from there?