The Transfer Deficit

The Most Expensive Assumption in HR

The global corporate training market is estimated at $394.5 billion a year (Training Industry, Inc., 2025 estimate).

It is worth being precise about what that number is: a market-sizing estimate of what is spent with and inside the training industry, not an audited total of corporate L&D budgets. We are not aware of any audited total — which is itself an early sign of the problem this report is about.

What the Money Buys Is Hard to Establish

What that money buys is harder to establish than it should be. The measures most organisations collect — completion, satisfaction and recall — do not by themselves establish that behaviour changed in work. Whether it changed, and for whom, is usually not recorded at all.

Several transfer-rate figures circulate widely in this field. We have not been able to trace the most common of them to a primary study that supports the exact denominator and time point they are quoted with, so this report does not use any of them.

The Problem Is Not New

One thing should be said before anything else: the transfer problem is not new. Anyone who has worked in L&D will recognise it. Baldwin and Ford reviewed it in 1988, and researchers have kept measuring it since.

What this report adds is not the finding. It is a question asked one step upstream — not why transfer fails, which is well studied, but why so little of the accountability structure that governs comparable budget lines has ever been built to answer for it.

Where This Report Looks Instead

The conventional response has been to redesign the training — better content, better platforms, better facilitation. This report argues that part of the problem lies elsewhere.

Before an intervention is designed, organisations rarely examine a variable that shapes whether development converts into behaviour: the individual’s readiness and rate of conversion — whether this person, at this moment, is converting what they are given into changed behaviour.

In VITALIS, the AERTHEIX instrument, this is read across two separate axes — the Growth Engine and Core Vitality — rather than as one variable.

A $394.5 Billion Line Item With No Attribution Model

Budget lines of this size are usually engineered for accountability. This one is not.

In aggregate, corporate learning is one of the largest discretionary people investments on earth. Budget lines of comparable size are normally built with an accountability chain attached, and this one generally is not.

The measures the function is asked for are not the measures that would settle the question.

ATD’s own 2026 State of the Industry report makes the point without any help from us: fewer than a quarter of organisations measure whether their training programmes achieve their organisational goals, and fewer still measure return on investment.

The measurement gap is not cosmetic. Research on the dominant evaluation habit — the post-course satisfaction survey — finds only a weak relationship between how much participants liked a programme and whether their behaviour actually changed (Alliger et al., 1997; Sitzmann et al., 2008).

For decades, the function has been asked to prove its value using instruments that were never designed to measure the destination. The issue is not that L&D chose weak measurement — it is that the operating model never demanded, funded or built the measurement infrastructure that transfer accountability requires.

Where the Investment Leaks

Transfer failure is not a single event. It is a two-stage leak: some of what is learned is never attempted in the work at all, and some of what is attempted stops being done within the following year. Organisations rarely measure either stage, which is why the size of each is contested and why this report does not attach a figure to them.

Not All Skills Transfer the Same Way

Research distinguishes closed skills, reproduced as trained, from open skills such as leadership, collaboration and judgement, applied flexibly. Trainee motivation and the work environment are more strongly related to transfer for open skills than for closed ones (Blume et al., 2010).

Bad Training Is the Instinctive Explanation. It Is Not the Best-Supported One.

The instinctive reading of low transfer is that the training was poorly designed.

Four decades of research point in a wider direction: whether learning is used depends on the person’s readiness before the training and on what their environment does after it, as well as on the design of the training itself.

Three Scenarios Most HR Leaders Will Recognise

The scenarios below are illustrations, not research findings.

A The high performer who plateaus

A regional sales manager consistently exceeds targets. The organisation invests in an executive leadership programme to prepare her for a GM role. She completes it, scores well, receives strong feedback. Twelve months later her leadership style is unchanged. The organisation concludes the programme was ineffective.

B The manager who knows but doesn’t change

A middle manager receives 360° feedback that his team feels micromanaged. He gets coaching, a delegation workshop, a mentor. He articulates the concepts perfectly in follow-up conversations. His team’s engagement scores never move. He leaves two years later, fundamentally unchanged.

C The new hire who doesn’t stick

A retail company invests heavily in structured onboarding for frontline staff. Content is well designed, facilitators experienced, managers briefed. Six-month attrition stays at 35%. Exit interviews cite “lack of growth opportunity” — the exact gap the programme was built to close.

In each case the organisation’s response was to improve the training. What went unexamined in all three was whether the person was ready, and able, to use what they were being given.

Each of these three scenarios has a name, and the names are not interchangeable. Part 4 sets out five patterns of momentum breakdown; the three cases above map onto three different ones, and the mapping is the whole point — because the correct response to each is different, and in every case above the organisation applied the response that fits only a CAPABILITY GAP.

Note the diagnostic honesty required in case B, which could sit in either of two patterns. The harder boundary is between PRIORITY GAP and RECOGNITION GAP, which look almost identical from the outside; the demonstration later in this report is built around exactly that distinction.

Diagnose the What, Ignore the Why

Competency frameworks answer a real question. It is not the question that determines whether development lands.

When development fails to deliver, the instinct is to sharpen the diagnosis of what is missing. This has produced increasingly sophisticated competency frameworks, gap analyses and skills taxonomies. These tools have real value — common language, role clarity, structured talent conversations.

But they tend to share one design assumption, so embedded it is rarely said out loud:

Once a person knows what they need to develop, and is given the means to develop it, they will.

This assumption holds for some people and fails for others, and the framework provides no tools for telling the two apart.

The Inversion: Where Budget Goes vs Where Learning Happens

The 70-20-10 model, usually attributed to Lombardo and Eichinger (1996) and drawing on Center for Creative Leadership research summarised in McCall, Lombardo and Morrison (1988), should not be read as a universal empirical law — it is an experiential observation for which we are not aware of a controlled replication, and its precise ratios have been challenged (Clardy, 2018).

Nevertheless, the practical imbalance it highlights is broadly supported by evidence: organisations tend to overfund formal learning assets while underfunding the application conditions, manager reinforcement and work-based practice that research most strongly links to durable behaviour change.

The commonly cited shape of the imbalance is this: formal training absorbs the majority of the L&D budget while being associated with a small share of durable learning, and on-the-job application and practice show the reverse.

We give the shape rather than precise percentages deliberately. The often-quoted 62%/10% and 70%/14% pairings circulate widely but combine two different sources measuring two different things — a budget-allocation survey and an experiential estimate — and we could not verify either pairing against a single primary document. The direction of the imbalance is well supported; the decimal places are not.

The imbalance persists for a procurement reason, not a pedagogical one.

Why does the inversion persist? Partly because classroom output is visible and easy to report. But the deeper reason is structural: formal training is purchasable — it has vendors, invoices and completion certificates. On-the-job design and manager reinforcement have no PO line; they cost management attention, which sits outside the L&D budget.

L&D has often been asked to prove value using measures that were never designed to capture transfer — and to deliver results through the only channel the procurement system recognises. The function did not choose this operating model; it inherited it. Changing the model requires more than better measurement — it requires a different conversation about what L&D is accountable for, and what it needs from the rest of the organisation to deliver.

The Hidden Variable: Readiness and Rate of Conversion

When two people receive the same intervention and only one changes, the difference may sit in readiness, opportunity, feedback, load — or the intervention itself.

Consider two employees who receive identical 360° feedback: both need to manage conflict more constructively. Both attend the same workshop. Three months later one has visibly changed; the other has not. The programme was the same for both. Something else differed, and part of it may have walked in with them.

What builds — and erodes — momentum

In the working description used here, high-momentum environments are those where feedback arrives close to the behaviour; development is framed as investment, not deficiency; people see evidence that growth leads somewhere; and trying something new is not career-threatening.

In low-momentum environments, development tends to be treated as remediation, feedback is annual and tied to ratings, high performers have no cognitive space left, and the implicit message is: perform the current role at maximum capacity — there is no time or recognition for growth. The same programme can land in these two environments as two very different interventions.

Five Patterns of Momentum Breakdown

Five working patterns, drawn from practice rather than from validation research, describe most of the cases we have seen where development does not convert. They are an AERTHEIX working framework, not a validated diagnostic classification, and they are offered as a way to structure a conversation rather than as a test that produces a result.

Each one calls for a different response, and a competency gap analysis is not designed to distinguish between them — which may be why the same programme, delivered well, can produce very different outcomes in the same cohort.

These five names are used consistently throughout this report and in everything that follows from it. Where you see them elsewhere in the series, they mean exactly what they mean here.

1 “I want to change, but I don’t know how.” — CAPABILITY GAP

Genuinely motivated, lacks a pathway. Structured guidance, worked examples and coaching land well — this is where standard training actually works.

2 “I understand, but I don’t think it matters.” — PRIORITY GAP

Feedback absorbed but not connected to anything personally valued. Investment is premature: a reframing conversation must come first.

3 “I disagree with the assessment.” — RECOGNITION GAP

Does not accept the premise. Training before this is resolved tends to produce resistance, compliance theatre or quiet disengagement.

4 “I agree, but the environment won’t let me.” — SYSTEM GAP

Motivated and self-aware, but the system inhibits the very behaviour being asked for. The constraint is structural — individual coaching can redirect blame and damage retention.

5 “I don’t have the bandwidth right now.” — CAPACITY GAP

Sometimes a genuine resource limit, sometimes a deflection. Distinguishing the two takes an honest conversation about timing and workload — not a new programme.

An experienced HR business partner may be able to tell these patterns apart, provided the question is asked before the intervention is designed rather than after it has failed.

The conversation is short. What it changes is which of five different responses gets funded.

When the Pattern Is Misread: Common Responses vs Effective Responses

What a Momentum Diagnosis Actually Looks Like

The five patterns are only useful if you can tell which one you are looking at. The following is a single anonymised case, reconstructed from a frontline retail environment. Identifying details have been removed and the timeline generalised; the dialogue is reported in substance rather than verbatim.

The situation. An experienced frontline employee had, in one year, recorded a strong result on an independent third-party service assessment — comfortably inside its upper band. Over the following period the organisation introduced a revised set of service standards. He declined to adopt them. Across two consecutive review cycles his assessment scores fell out of the top band and eventually sat near the bottom of it.

What was actually observable. Not an attitude — two behaviours. He did not take part in the newly introduced structured floor walks, and he stopped attending team briefings. Both were scheduled, both were logged, and both were verifiable without anyone having to interpret his state of mind.

What he said when asked. In substance: I understand your approach is sound, but we work brand-first. I have already done what the brand asks. The assessors mark me down because they do not understand the brand's own requirements — so a low score is a problem with the guidelines, not with me. I can hardly abandon the brand standard and just make small talk instead.

What the follow-up produced. His line manager had pursued the same issue repeatedly. After several rounds the manager's own summary was that there was nothing further to be done — the person simply did not want to change. That sentence is worth pausing on, because a manager reporting exhaustion is a diagnostic signal in its own right, not merely a management failure.

What was expected, and why it was wrong. The initial reading was a SYSTEM GAP — the pattern where a motivated person is blocked by conditions they do not control, which is the one most often missed and the one being actively looked for. The evidence did not support it. Nothing in his environment prevented him from attending a briefing; the briefings were scheduled, and his colleagues attended them.

The mis-read is recorded deliberately: a framework that lets you find whatever you went looking for is not a diagnostic instrument, and the only way a reader can tell the difference is if the cases that overturned the initial reading are published alongside the ones that confirmed it.

What it actually reads as. On the surface this looks like a PRIORITY GAP — he appears to be saying the new standard is not what matters to him. It is not.

Read the sentences again: he is not deprioritising the assessment, he is rejecting its validity. The assessors are wrong; the guidelines are the problem; the low score is evidence of someone else's error. That is a RECOGNITION GAP, and the brand-first argument is not a competing priority but a legitimising frame for refusing the premise.

The test that separates the two is simple. Ask whether the person accepts that the gap exists. Someone in a PRIORITY GAP says yes, and I have bigger things to worry about. Someone in a RECOGNITION GAP says no, and here is why the measurement is wrong.

Why the distinction cost real money. The organisation responded as if it were a CAPABILITY GAP: more training, more standards, more briefings. Part 4's table describes what tends to happen when you train against a premise the person has not accepted — resistance, compliance theatre, or quiet disengagement. All three were visible at one point or another. Roughly eighteen months of declining third-party scores followed, in a role where those scores feed directly into commercial judgement.

What eventually moved him — and what did not. It does not appear to have been the training, the coaching or the manager. What appears to have changed the behaviour was a change in his reference group: new colleagues joined the department who engaged with the new standards readily, the visible distance between them and him widened, and he began to register that he was being left outside. He resumed attending briefings and training.

The part that must not be rounded off. The RECOGNITION GAP was never actually closed. He still maintains that the standards keep shifting. What changed was the social cost of holding that position, not the position itself.

That distinction matters commercially: behaviour bought by peer pressure reverts when the peer group changes, and a compliance recovery is not the same asset as a genuine one.

A correct diagnosis at the start would not have avoided the disagreement. It would have addressed it directly — feedback alignment, evidence he could examine himself, a conversation about the measurement before a conversation about the behaviour — and might have cost less than eighteen months.

Telling the Patterns Apart: One Question Per Boundary

The five patterns are easy to name and hard to separate, and the pairs that get confused tend to be the same pairs. In practice a diagnosis rarely fails because someone did not know the five categories. It fails because two of them present almost identically in the room, and the wrong one is more comfortable to conclude.

Each boundary can often be tested with a single question, asked directly.

What Four Decades of Research Actually Says

Four decades of research point to interacting person, design and environment conditions; practice often still overweights the course itself.

BALDWIN & FORD (1988) — the foundational transfer model: outcomes are a function of learner characteristics, training design, and work environment. Thirty-eight years later, much of the spend still targets design, while the other two variables — the person and the environment — are frequently left undiagnosed.

Self-Determination Theory: Not All Motivation Is Equal

Controlled motivation — driven by external pressure: compliance targets, fear of consequence, “because HR said so.” Tends to produce short-lived, monitored behaviour that may rebound when oversight relaxes.

Autonomous motivation — driven by internalised value: the person sees the change as theirs. The motivation type most consistently linked in research to durable behaviour change (Deci & Ryan, 2000; Gagné & Deci, 2005). A programme cannot install it directly; conditions can make it more likely.

THE LTSI — the Learning Transfer System Inventory (Holton, Bates & Ruona, 2000) is a validated 16-factor transfer system scale. Its factors span the person, the training design and the work environment; several of them describe individual states such as readiness, motivation to transfer, self-efficacy and outcome expectancy. What matters for this report is that the instrument treats those states as part of the system at all, and that most development decisions are taken without asking about any of them.

Acknowledging the Field

Several established methods already work on this problem, and they work downstream of where this report is looking.

This report is not the first to address transfer failure. The Six Disciplines of Breakthrough Learning (Wick, Pollock & Jefferson, 2006) made a significant contribution by extending accountability beyond the classroom into application and reinforcement. Brinkerhoff’s Success Case Method (2003) introduced a rigorous approach to identifying what actually transferred. The New World Kirkpatrick Model (Kirkpatrick & Kirkpatrick, 2016) updated the four-level evaluation model for use in a business setting.

These methods advanced the field materially, particularly in post-training reinforcement. However, they generally start from an assumption: that the training has been correctly targeted at the right person, at the right time.

This report addresses what happens upstream of that assumption — the diagnostic step that determines whether a person is in a convertible state before any investment is designed. The five momentum patterns described in Part 4 are an attempt to make that upstream variable visible, specific and actionable.

The Cost of Misdiagnosis

Misdiagnosis is not merely wasteful. Two of the five patterns may be made worse by the standard response.

Send an employee in a RECOGNITION GAP to training, and the cost is not only the programme fee — it is also the cohort energy, the manager hours, and the reinforced narrative that development is theatre.

Coach an employee in a SYSTEM GAP individually, and the problem is not solved but redirected — from a structural issue to a personal one. The employee may absorb blame for a situation they cannot control, and engagement and retention can pay the price.

Whether a programme pays back may depend at least as much on the readiness of the people in the room as on the programme itself.

A Different Starting Point

Nothing here argues for abandoning what already works. It argues for one step being added in front of it.

The implication is not that competency frameworks should be abandoned — they create common language and structure that serious talent work needs. The implication is that gap analysis should be preceded by a momentum diagnosis. Before any individual development plan is funded, three questions deserve honest answers:

**Q1 **Does this person currently have the motivation to use this opportunity — connected to something they personally want to achieve?

**Q2 **Does their environment support the behaviour change being asked for? If the constraint is structural, individual investment is unlikely to move the needle.

**Q3 **What is the specific nature of their readiness or resistance — and which of the five patterns does it match: CAPABILITY, PRIORITY, RECOGNITION, SYSTEM, or CAPACITY?

Self-Diagnostic — Where Is the Momentum in Your Team?

Think of one person who completed a development programme in the last twelve months without visible behaviour change:

VITALIS is not a training-ROI instrument and does not establish why transfer failed. It creates a structured record of whether behaviour changed in real work, what evidence supports that reading, and where the explanation remains open.

That record can help a manager decide whether the next move is instruction, coaching, a change in conditions, a change in load, or more observation. Founding deployments will test whether the protocol can make those distinctions consistently and usefully, without overstating what the evidence shows.