The short answer: Good training programs fail after delivery, not during it. Research tracking trainees found that 38% of training content is not applied immediately after a program, rising to 66% after a year. What decides the outcome sits outside the program itself: an expectations conversation before it, a real application task after it, and a manager who asks.
Three weeks after a program ends, ask a participant what they now do differently. In most organizations the honest answer is "nothing yet." The more revealing fact is that nobody had asked before you did — which is exactly why nobody noticed.
You can commission the best training program on the market — built on your actual role tasks, written in language people understand on first read, tied to a measurable competency — and still watch it produce nothing. The fault is not in the program. It is in how the organization adopts it.
This article is about the part that starts once the design is signed off: why training programs fail even when they are well built, which everyday behaviours dismantle them, and how to construct a training journey that makes application a predictable outcome rather than a happy accident.
If you haven't reached the design stage yet, the previous article covers the four decisions taken before the first slide is written: how to design a training program that actually changes performance.
Why training programs fail: most of the impact sits outside the program
One number circulates in this industry more than any other: that only 10% of training transfers to the job. It is not a research finding. It originated as a rhetorical device — David Georgenson opened a 1982 article by asking readers how often they had heard training directors estimate that figure. He cited no data, because he was not reporting any. Four decades of citation turned a conversational aside into an accepted fact.
The real picture is less catastrophic and far more useful. Studies that tracked trainees after programs found that 38% of training content is not applied immediately after delivery, rising to 56% at six months and 66% at twelve (European Journal of Training and Development, 2024). Measured differently — at organization level rather than per trainee — benchmark research put "scrap learning," training that is delivered but never applied, at 45% of everything organizations deliver (CEB / KnowledgeAdvisors, 2014).
Look at the shape of those numbers rather than their size. The figure gets worse over time. Whatever is going wrong is not going wrong during the learning. It is going wrong afterwards.
Nothing in a completion report tells you whether a single person does anything differently.
Professor Robert Brinkerhoff turned that observation into a model worth pinning to the wall of every L&D function: 40/20/40. The program itself — the content, the facilitator, the room, the platform — accounts for roughly 20% of the impact. The other 80% is split between what happens before it and what happens after (Brinkerhoff & Apking, High Impact Learning, 2001).
| Phase | Share of impact | What happens in it |
|---|---|---|
| Before the program | 40% | Selection, expectations, preparation |
| During the program | 20% | Content and delivery |
| After the program | 40% | Application, follow-up, measurement |
That reframes the question. Instead of "is the program any good?", the question becomes: what did we set up before it, and what did we commit to after it?
The everyday practices that dismantle a training program
Nobody decides to sabotage a training program. It happens through small, routine behaviours, each of which has a defensible operational logic in the moment, and which together neutralise the program before it begins.
What managers do
| The practice | What it does to the program |
|---|---|
| Nominating whoever can be spared this week rather than whoever has the gap | Selection becomes a scheduling decision. The person in the room doesn't need the skill; the person who needs it stays at their desk. |
| Approving attendance, then calling the participant out mid-session for something urgent | They leave and return twice. The message the whole team receives: this ranks last, so treat it accordingly. |
| Never finding out what the program actually covers | A manager cannot ask for a behaviour they don't know exists, or notice its absence. Application becomes optional by default. |
| Opening the first conversation after the program with the backlog | The program is buried in ten minutes. The employee concludes that what they learned is extra, not part of the job. |
| Sending someone on training after a mistake | Training becomes a disciplinary instrument, and nomination for the next program becomes something people quietly avoid. |
| Modelling the opposite of what the team was trained on | A program on listening to customers, run by a manager who interrupts in every meeting. Observed behaviour beats documented behaviour every time. |
| Signing the post-training follow-up form without reading it | The cycle closes administratively and fails in reality — then appears in the annual report as a success. |
What employees do
| The practice | What it does to the program |
|---|---|
| Running the course on a second monitor while working | Content plays; nothing is read. Completion rises, capability doesn't, and the dashboard looks healthy while performance stays flat. |
| Sharing assessment answers in the team chat | The post-assessment loses all diagnostic value, and L&D loses its only instrument for spotting which module didn't land. |
| Attending because the certificate is required for promotion | The goal becomes clearing a requirement, not acquiring a capability. The employee isn't at fault here — they're responding accurately to what the organization actually rewards. |
| Trying the new behaviour once, hitting friction, reverting | The most common failure mode and the least visible. The attempt happened. Nobody was there to back it in the first fortnight. |
| Deferring application until "after month-end" | The quiet window never arrives. Thirty days later the operational detail has evaporated and only a vague impression remains. |
Note what all twelve rows have in common: not one of them is a content problem. The program can be excellent in every single case.
Four conditions any new behaviour needs to survive
Donald Kirkpatrick identified four conditions without which a new behaviour will not persist after training. The New World Kirkpatrick model later formalised them as required drivers — the reinforcement, monitoring, accountability and reward that convert learning into behaviour. All four can be tested cheaply before a program launches.
| Condition | What it means in practice | Who owns it |
|---|---|---|
| Desire to change | The employee can see that the new behaviour serves them, not a report | The line manager |
| Knowing what and how | They know the specific operational step, not the general principle | The training program |
| A climate that permits it | The tool, the permission and the procedure allow it — and someone backs the first attempt | The organization |
| Reward for applying it | People who apply it are noticed, and the metrics don't reward the opposite | The performance system |
The last two are what sink most programs. A pattern we see repeatedly: a bank trains its relationship managers on consultative discovery — ask before pitching, log what you learn, come back with something specific. The program is genuinely good and the scenarios come from real accounts.
Then the CRM's call-logging screen requires nine mandatory fields and takes four minutes. And the RM's weekly target is contacts made, not conversations that went anywhere. So the behaviour that gets rewarded is the fast, shallow call — the exact behaviour the program was built to replace.
The team learned one thing and was paid for the opposite. None of this shows up in the training report, because the training report measures completion, not permissions and targets.
The training journey: what happens before, during and after
A training journey is the sequence of specific commitments that surround a program — with named owners and dates attached — so that application doesn't depend on individual enthusiasm. It is the difference between a program that gets delivered and a program that gets used.
This is not a theory. In a field study of 91 trainees at a Fortune 200 company, participants were randomly assigned to two groups: one whose managers held a pre-training expectations discussion and a post-training follow-up discussion, and one whose managers did not. The first group reported significantly higher application, and saw their work environment as more supportive of applying what they'd learned (Brinkerhoff & Montesino, Human Resource Development Quarterly, 1995).
Two conversations. That was the entire intervention.
Before the program — 3 weeks out (40% of impact)
- Nomination based on a documented gap in the competency framework, not on who is available.
- A 15-minute expectations conversation: why you were nominated, and which task you will do differently afterwards.
- A pre-assessment on the target competency.
- A real application opportunity booked into the participant's calendar within three weeks of the program ending.
During the program — delivery days (20% of impact)
- Time that is genuinely protected — no calls, no attending "from the desk."
- The participant leaves with a written application plan of one or two specific commitments, not ten intentions.
- Each commitment phrased as an observable behaviour: "I use the escalation form on every case past 24 hours," not "I improve my communication."
After the program — 90 days (40% of impact)
- Days 1–14: the booked application task is carried out, and the manager observes it directly and comments.
- Day 30: two questions in the one-to-one — where did you apply it, and what got in the way. The second matters more than the first.
- Days 60–90: post-assessment on the same competency, a behavioural observation logged in the performance system, and a read of the operational metric the program was built to move.
Assigning ownership is half the battle, because each party routinely assumes another one is handling the follow-up:
| Party | Specific responsibility |
|---|---|
| L&D | Designing the program against the gap, preparing the pre- and post-assessment, and giving managers a one-page conversation guide before launch. |
| Line manager | Nomination, the expectations conversation, protecting the time, booking the application task, the day-30 follow-up, and the behavioural observation. |
| Employee | A two-item application plan, carrying out the booked task, and naming the obstacle instead of quietly working around it. |
| Executive team | Clearing the operational blocker when it is raised: the permission, the procedure, and the metric that rewards the opposite behaviour. |
Five early signals that a program is heading for failure
You don't need to wait for the year-end review to know the answer. The reasons why training programs fail show up early, and every one of them is fixable at the moment it appears:
- Before launch: managers cannot state, in one specific sentence, what will change in their team's work after the program.
- At nomination: the names came from availability or from spreading places fairly across departments, not from a documented gap.
- At scheduling: not a single application task is booked in anyone's calendar for after the program.
- Day one back: the first question the participant hears is about the backlog, and the program isn't mentioned.
- Day thirty: the performance system contains no trace of the program — no goal, no behavioural observation, no post-assessment.
Signal one alone justifies pushing the launch back two weeks. That delay is far cheaper than running a full program nobody applies.
"But our managers don't have time for any of this"
This is the most common objection, and it deserves an answer in numbers rather than in encouragement.
Add up what the journey actually asks of a line manager per participant, per program: 15 minutes for the expectations conversation, 15 minutes for the day-30 follow-up, 15 minutes to observe and log at day 90. Forty-five minutes, spread across three months.
The objection also has a legitimate side. Line managers genuinely are overloaded. LinkedIn's 2025 Workplace Learning Report found that 50% of respondents said managers lack the support they need to back their people's development, and that only 15% of employees said their manager had helped them build a career plan in the previous six months — five points down on the year before (LinkedIn Workplace Learning Report, 2025).
But the conclusion that follows is not to drop the follow-up. If an organization cannot find 45 minutes per participant, it also cannot afford two days of their time or the cost of the program.
When time is short, run fewer programs. Don't run the same number with less follow-up. Two programs that get applied beat six that get delivered.
And that is a decision for L&D more than for managers. An overloaded annual training calendar is usually the reason none of it can be followed up properly.
Where Lumofy fits
Lumofy is a workforce capability platform that also produces custom eLearning content for organisations across the Gulf, delivering large course libraries in Arabic and English on compressed timelines. What matters for this article is that the four elements deciding application live inside the same platform rather than in separate files:
- The competency framework settles nomination. A standard is defined per role and employees are measured against it, so the person nominated is the one with the gap, not the one who is free.
- Pre- and post-assessment on the same competency. The program's result becomes a number you can put in front of the executive team, and the modules that moved nothing surface before the next cohort.
- Learning pathways tied to the individual development plan inside the performance cycle. What someone learned appears in front of their manager in the same place they discuss goals — not in a separate system nobody opens.
- Managers see their team and their competency levels on one screen, which turns the day-30 follow-up from an extra task into part of a meeting they were already having.
Organizations that work in this order report identifying role skill gaps and high-potential talent up to 40% faster than with their previous approach, and stronger alignment between training and business goals. Worked examples are in the Lumofy customer stories.
Frequently asked questions
Because roughly 80% of a program's impact is created outside it — in nomination and the expectations conversation beforehand, and in application and follow-up afterwards. An excellent program satisfies one of the four conditions for behaviour change: knowing what and how. Desire, climate and reward are set by the organization, the line manager and the performance system.
Training transfer is the extent to which people apply what they learned once they are back doing the job. It decays because a new behaviour has to compete with an established one under time pressure. Research found 38% of content not applied immediately after a program, 56% at six months and 66% at twelve — the gap widens without reinforcement.
Three specific moments: an expectations conversation before the program explaining why the person was nominated and what will change in their work, a real application task booked for afterwards, and a day-30 follow-up with two questions — where did you apply it, and what got in the way. A 1995 randomised field study found those two conversations alone significantly raised application.
Three pieces of evidence together: a post-assessment on the same competency measured before the program, a behavioural observation logged by the line manager within 90 days, and movement in the operational metric the program was built to shift. Completion rate is not evidence of application — it measures that content was opened, not that behaviour changed.
When it turns out the problem was never a capability gap. If an employee could do the task but the tool, the permission or the metric prevents it, no program will change anything however well built. Fix the operational blocker first, then revisit whether training is needed at all.
Where to start
Take the last training program your organization ran, and look in the performance system for a single trace of it thirty days later: a goal linked to it, a behavioural observation, a post-assessment.
If there is nothing, the program didn't fail. It never started.
