For something that has shaped billions of dollars of learning investment globally, the 70-20-10 model rests on surprisingly fragile foundations.
It is quoted in boardrooms, embedded into learning strategies, and taught as near doctrine in leadership development circles. It has influenced how organisations allocate budgets, design journeys, and justify deprioritising formal learning. It has become, in many ways, one of the most convenient truths the learning profession has chosen to believe.
And yet, when one steps back and examines its origins with rigour, a very different picture emerges.
The model was never scientifically validated. It was never experimentally tested. It was never intended to be prescriptive.
What it was, at its core, was a retrospective reflection by a small group of successful executives attempting to explain how they believed they had learned over the course of their careers.
The leap from perception to principle happened quietly. And that is where the problem begins.
From Insight to Orthodoxy: How a Heuristic Became a Rule
The origins of the 70-20-10 model trace back to research conducted in the 1980s, where approximately 200 executives were asked to reflect on how they developed critical capabilities. Their responses were categorised into three broad sources: challenging experiences, developmental relationships, and formal education.
The now-famous ratio emerged from this exercise. Not from measurement, but from memory. Not from controlled study, but from interpretation. This distinction matters more than we often acknowledge.
Because what executives recall about their learning journeys is not the same as how learning actually occurs. Human memory is selective, narrative-driven, and influenced by hindsight bias. Individuals tend to attribute growth to visible, high-impact experiences while underestimating the invisible scaffolding that made those experiences meaningful in the first place.
Formal learning is rarely remembered as transformative in isolation, but it often provides the language, frameworks, and mental models that enable individuals to make sense of experience later.
To convert retrospective perception into a prescriptive ratio is, therefore, not just a stretch. It is a distortion.
The Illusion of Evidence
Despite its widespread adoption, there is no robust empirical evidence that validates the 70-20-10 distribution.
Multiple reviews, including those referenced by the Association for Talent Development and independent academic analyses, have consistently pointed out that the model lacks controlled experimental backing. Studies that examine learning effectiveness do support the integration of experiential and social learning, but they do not confirm any fixed proportion.
This is a critical distinction.
The idea that people learn through experience and interaction is well supported by research from scholars such as David Kolb and Albert Bandura. Experiential learning theory and social learning theory provide strong conceptual foundations for why learning extends beyond the classroom.
But the leap from “these sources matter” to “these sources contribute in fixed percentages” is not supported by evidence. The numbers are not just approximate. They are arbitrary. And yet, organisations continue to treat them as precise.
The Danger of Misinterpretation
Perhaps the most significant damage caused by the 70-20-10 model is not theoretical, but practical.
In many organisations, the model has been interpreted as justification to reduce investment in formal learning under the assumption that most learning will happen naturally through work. The logic appears sound on the surface. If 70 percent of learning happens on the job, then resources should be directed toward creating experiences rather than designing structured learning.
But this interpretation rests on a flawed assumption. It assumes that experience automatically leads to learning.
Research on expertise, particularly the work of Anders Ericsson, suggests otherwise. High-quality learning requires deliberate practice, structured feedback, and opportunities for reflection. Without these conditions, individuals do not necessarily improve through experience. They simply repeat existing patterns.
In fast-paced organisational environments, where time for reflection is limited and feedback is inconsistent, experience can just as easily reinforce ineffective behaviour as it can build capability.
When organisations rely on experience without design, they are not accelerating learning. They are leaving it to chance.
A Model Built for a Different World
Even if one were to set aside questions of empirical validity, the relevance of the 70-20-10 model must be examined in the context of how work itself has changed.
The model emerged in a time when roles were relatively stable, career paths were linear, and individuals had the opportunity to build capability gradually over time. Learning on the job worked because the job itself provided continuity.
That continuity no longer exists.
Today’s workplace is characterised by rapid technological change, evolving business models, and increasing complexity of roles. Individuals are often expected to operate in ambiguity, make decisions without precedent, and lead in contexts for which there are no established playbooks.
In such environments, the assumption that capability will emerge organically from experience becomes increasingly untenable.
Experience can deepen capability, but it cannot substitute for it.
When individuals are placed into roles without the foundational capability required to navigate them, experience does not necessarily lead to growth. It often leads to anxiety, avoidance, or over-reliance on familiar behaviours that may no longer be effective.
The modern workplace demands preparation before exposure, not the other way around.
The Capabilities That Experience Cannot Build Alone
There is another, more fundamental shift that challenges the relevance of the model.
The capabilities that define effectiveness today are not primarily technical. They are human, cognitive, and behavioural.
The ability to think systemically, to navigate ambiguity, to exercise judgment under uncertainty, to build trust in distributed teams, to lead without authority, and to learn and unlearn continuously are not capabilities that emerge automatically through exposure.
They require intentional development. They require safe spaces for practice.
They require feedback loops that allow individuals to refine behaviour before consequences become real.
To assume that these capabilities will develop through unstructured experience is to misunderstand how complex human capability is built.
From Learning Distribution to Capability Design
What this suggests is not that the principles underlying 70-20-10 are entirely invalid, but that the way they have been interpreted is deeply flawed.
The future of learning does not lie in distributing effort across percentages.
It lies in designing capability systems. This requires a shift in mindset from asking, “Where will learning happen?” to asking, “How will capability be built?”
It requires recognising that formal learning must do more than introduce concepts. It must create the initial conditions for capability through practice, simulation, and feedback. It requires structuring social learning rather than assuming it will occur organically. It requires curating experiences so that they stretch capability without overwhelming it.
Most importantly, it requires sequencing. Capability must be built before it is tested. Experience must reinforce, not replace, development.
The Cost of Holding On
The continued reliance on the 70-20-10 model is not a harmless legacy. It carries consequences.
It allows organisations to underinvest in structured capability building while overestimating the learning that will emerge from work itself. It creates the illusion that learning is happening because activity is visible. It shifts responsibility for development onto individuals without equipping them with the tools required to grow.
And perhaps most critically, it prevents organisations from asking a more difficult question.
Not how learning is distributed, but whether it is working. Because in the end, capability is not measured by participation, exposure, or even experience.
It is measured by behaviour. By decisions made under pressure. By actions taken when the answer is not obvious. By the ability to operate effectively in environments that do not resemble the past.
Reframing the Future
The learning organisations of the future will not be those that adhere most closely to inherited models.
They will be those that are willing to question them. They will recognise that learning is not an organic by-product of work, but a system that must be deliberately designed. They will move beyond heuristics and begin to build architectures that align learning with capability, and capability with business outcomes.
They will understand that the real shift is not from 10 to 70. It is from assumption to intention. Because the most dangerous models are not the ones that are wrong. They are the ones that feel right, are widely accepted, and go unquestioned for too long.
And 70-20-10 may well be one of them.
