Every factual claim below is cited, and every citation was opened and read before it was used. That is not a flourish. It is the argument.
The number everyone knows
You have heard it in a keynote, seen it in a vendor deck, or read it in the opening paragraph of a peer-reviewed paper: only 10% of training transfers to the job. Often it carries a dollar sign. American organizations spend roughly $100 billion a year on training, the line goes, and no more than a tenth of that produces any change in behavior at work.
The figure is compact, memorable, and unflattering enough to sound like hard truth. It has justified budget cuts and framed four decades of argument about why corporate learning underdelivers. It is repeated by the people who build training and by the people who want to defund it — usually a sign that a number has stopped being evidence and become furniture.
It has no evidentiary basis. Not a weak one, not a dated one. None. And the trail back to its origin is short enough that any of us could have walked it at any point in the last twenty-five years.
What the 1982 article actually says
The citation almost always resolves to the same place: Georgenson, D. L. (1982), “The Problem of Transfer Calls for Partnership,” Training and Development Journal, 36(10), 75–78. Georgenson was then manager of product development at Xerox Learning Systems. The article runs four pages and is still available from ATD, the journal’s publisher. We read it. It opens like this:
“How many times have you heard training directors say: ‘I need to find a way to assure that what I teach in the classroom is effectively used on the job’?
‘I would estimate that only 10 percent of content which is presented in the classroom is reflected in behavioral change on the job. With increased demand from my management to demonstrate the effectiveness of training, I’ve got to find a way to deal with the issue of transfer.’” (Georgenson, 1982, p. 75)
That is the entire basis of the claim. Georgenson is not reporting a study, a survey, or a benchmark. He is doing what practitioner writers have always done in a first paragraph: putting a plausible complaint into the mouth of a composite training director to establish that the topic is worth reading about. He cites no source for the 10% figure because there was none to cite.
The rest is a sensible, unremarkable piece on manager support — set the stage before the course, tie it to a real work project, give timely feedback, build the content into the unit’s daily work. It is a decent article. It deserved better than what happened to it.
One detail matters for what follows. Nowhere does Georgenson discuss the cost of training. There is no $100 billion in the piece — no dollar figure of any kind, in any paragraph. That number was added later and then attributed to him.
How a rhetorical question became a finding
Robert Fitzpatrick traced the propagation in 2001 in The Industrial-Organizational Psychologist, the newsletter of SIOP, under the title “The Strange Case of the Transfer of Training Estimate.” The chain he documented runs like this.
Wexley and Baldwin (1986) cited Georgenson for the estimate that “only 10% of the dollars spent on training results in actual behavioral change back on trainees’ jobs.” Dollars had entered the claim. Two years later, Baldwin and Ford’s landmark review in Personnel Psychology opened by stating that American industries spend up to $100 billion annually on training and that “not more than 10% of these expenditures actually result in transfer to the job” — citing Georgenson (1982). The $100 billion now had a citation. The citation did not lead anywhere.
From there it spread on momentum. Dickson and Bamford (1995) attributed it to “Georgenson (1981),” misdating a source they evidently had not opened. Facteau and colleagues (1995) upgraded it from an estimate to a “finding.” Broad and Newstrom’s widely used 1992 book quoted Baldwin and Ford rather than Georgenson, adding a layer of insulation between the claim and its missing evidence. Fitzpatrick set the project aside around 1996, assuming the outbreak had burned out — then opened the April 2001 Journal of Applied Psychology and found the claim in an introductory paragraph, cited to Georgenson (1982) twice: once for the $100 billion, once for the 10%.
Fifteen years, two manufactured facts, one rhetorical question. Not one of the citing authors appears to have opened the 1982 article and read its first page. The failure was not a bad study or a misread confidence interval — there was no study. The mechanism was social. Each citation made the next one safer, until the claim became something everyone knew, a category that does not require anyone to have checked.
Three corrections, and why they did not take
The field has been told. Repeatedly. Will Thalheimer reprinted Fitzpatrick’s piece with SIOP’s permission in 2010 and drew the obvious conclusion: ignore the number, and build the habit of checking the claims you inherit. Jeanne Farrington returned to it in 2011 in Performance Improvement Quarterly, under a title that makes the position plain: “Training transfer: Not the 10% solution.” In the same journal and year, Ford, Yelon, and Billington documented the field’s acceptance of the claim “despite the lack of empirical, behavioral evidence.”
Note who wrote that last one. J. Kevin Ford is the Ford of Baldwin and Ford (1988) — the review that gave the $100 billion its citation. The correction came from the source of the amplification. The claim kept circulating anyway.
What the evidence actually shows
Correcting a myth is only half the work. The more useful question is what replaces it — and the honest answer is that “what percentage of training transfers” is close to unanswerable as posed. Transfer varies by content, design, learner, manager, and time elapsed. A single headline percentage was never going to exist.
What does exist is better. Saks and Belcourt (2006) surveyed training professionals in 150 organizations and found reported application rates of 62% immediately after training, 44% at six months, and 34% at one year. Those are self-reports rather than observed behavior — but they describe a decay curve, not a flat 10% wall, and the shape of that curve is itself the actionable finding.
The strongest recent evidence is meta-analytic. Lacerenza and colleagues (2017), synthesizing 335 independent samples in the Journal of Applied Psychology, found leadership training produced positive effects across all four Kirkpatrick criteria: reactions (δ = .63), learning (δ = .73), transfer (δ = .82), and results (δ = .72). Transfer showed the largest effect of the four.
Read that carefully, because misusing it is the temptation that created the 10% myth in the first place. A δ of .82 does not mean 82% of training transfers. It is a corrected standardized mean difference — how far trained groups moved relative to untrained ones, not a proportion of content surviving the trip to the job. In practical terms that is roughly a 28% increase in leadership behaviors performed on the job. Substantial. Not magic. And not a percentage of anything transferring.
The moderator findings carry the practical value. Transfer improved when programs were built on a needs analysis, included feedback, used multiple delivery methods with practice among them, were spaced rather than massed, and were instructor-led face to face. Transfer is a design variable, not a fixed tax on learning — a materially different management conversation than the one the 10% figure produces.
The question this leaves for your L&D function
The uncomfortable part is not that the field believed something false. Every field does. It is how cheap the correction was. Georgenson’s article is four pages long, publicly available for decades, and the disconfirming evidence sits in its first paragraph. Checking cost one afternoon. Not checking cost four decades of an industry describing itself as 90% ineffective on the authority of an imaginary training director.
So the question is not whether your team repeats the 10% figure. It is what else is in circulation on the same terms. A short, uncomfortable audit:
- Which numbers appear in your decks, business cases, and program rationales without a named source?
- For the ones that carry a citation, has anyone opened the cited source — not the paper that cited it, the source itself?
- Which design standards rest on a model that was never validated, or was validated and later failed to replicate?
- When a claim flatters your position, do you check it as hard as one that threatens it?
We are not exempt. Consultancies have every incentive to repeat a statistic that makes the problem we sell against look enormous, and this was an unusually convenient one. Anyone who has used it — and many of us have — was propagating an unsourced claim in order to sell a solution.
The discipline this calls for is unglamorous and entirely learnable. Check the primary source before the claim leaves the building. Cite what you read, not what you read about. If the source turns out to be a rhetorical question from 1982, you want to be the one who found that out — not the next citation in the chain.
Sources
1. Georgenson, D. L. (1982). The problem of transfer calls for partnership. Training and Development Journal, 36(10), 75–78. Full text (ATD): https://assets.td.org/m/5943dbc6e4b0d42b/original/The-Problem-of-Transfer-Calls-for-Partnership.pdf
2. Fitzpatrick, R. (2001). The strange case of the transfer of training estimate. The Industrial-Organizational Psychologist, 39(2), 18–19. Society for Industrial and Organizational Psychology. Reprinted with permission at: https://www.worklearning.com/2010/11/01/georgenson/
3. Thalheimer, W. (2010, November 1). Georgenson [Blog post, including the full reprint of Fitzpatrick, 2001]. Work-Learning Research. https://www.worklearning.com/2010/11/01/georgenson/
4. Farrington, J. (2011). Training transfer: Not the 10% solution. Performance Improvement Quarterly, 24(1), 117–121. https://doi.org/10.1002/piq.20105
5. Ford, J. K., Yelon, S. L., & Billington, A. Q. (2011). How much is transferred from training to the job? The 10% delusion as a catalyst for thinking about transfer. Performance Improvement Quarterly, 24(2), 7–24. https://doi.org/10.1002/piq.20108 (ERIC record EJ940148)
6. Baldwin, T. T., & Ford, J. K. (1988). Transfer of training: A review and directions for future research. Personnel Psychology, 41(1), 63–105. https://doi.org/10.1111/j.1744-6570.1988.tb00632.x
7. Saks, A. M., & Belcourt, M. (2006). An investigation of training activities and transfer of training in organizations. Human Resource Management, 45(4), 629–648. https://doi.org/10.1002/hrm.20135
8. Lacerenza, C. N., Reyes, D. L., Marlow, S. L., Joseph, D. L., & Salas, E. (2017). Leadership training design, delivery, and implementation: A meta-analysis. Journal of Applied Psychology, 102(12), 1686–1718. https://doi.org/10.1037/apl0000241
9. Paul, M. (2021, September 15). Umbrella summary: Leadership training. Quality Improvement Center for Workforce Development. https://www.qic-wd.org/umbrella-summary/leadership-training (source for the practical-percentage translation of the Lacerenza et al. effect sizes)
10. Park, Y., & Jacobs, R. L. (2008). Transfer of training: Interventions to facilitate transfer of training based on time and role perspectives. Academy of Human Resource Development. https://files.eric.ed.gov/fulltext/ED501672.pdf (secondary confirmation of the Saks & Belcourt 62/44/34 figures)
Secondary citations named inside the Fitzpatrick (2001) chain — Wexley & Baldwin (1986); Alavi (1994); Dickson & Bamford (1995); Facteau, Dobbins, Russell, Ladd, & Kudisch (1995); Broad & Newstrom (1992); Robinson & Robinson (1995); Smith-Jentsch, Salas, & Brannick (2001) — are reported as Fitzpatrick documented them, with his page references. Full details appear in his reference list at the URL above.