The headlines fixate on office mandates and cost cuts while a quieter crisis deepens: only one in five employees worldwide report being engaged at work, a drop that Gallup estimates drains roughly 9 percent of global GDP in lost productivity, about 10 trillion dollars (Gallup, 2025). Manager engagement has fallen sharply too, from 27 percent in 2024 to 22 percent in 2025, even as organizations rely on managers to fix morale and skills (Gallup, 2025). More seminars will not reverse this. The problem sits in how work is designed. Self-directed education is not a perk or a platform. It is an operating system embedded in the everyday.
What the data says about the learning gap
Gallup’s latest global survey is blunt about the scale and cost of disengagement, and equally clear that some organizations do far better, with manager engagement at 79 percent in best-practice firms compared with a 22 percent global average (Gallup, 2025). That spread implies a design problem, not a fate problem. If the conditions of work throttle initiative, no catalog of courses will create energy. Learning follows from autonomy, capability, and connection. That is the logic of self-determination theory, which has increasingly shaped how educators and managers try to build intrinsic motivation.
A recent meta-analysis catalogues what, until now, had been scattered intervention work in education. Wang and colleagues synthesized 36 full-text studies with 11,976 participants, selected via a PRISMA process from 505 records across nine databases, and, critically, assessed interventions across all three basic psychological needs: autonomy, competence, and relatedness (Wang et al., 2024). It is the first systematic review in more than a decade to take this full-scope view; an earlier review examined autonomy support alone. The point for executives is not to memorize theory. It is to see that the field has moved from slogans to structured tests of how to create the conditions that drive people to engage of their own accord. Whether in a school or a factory, those conditions can be designed.
Safety before syllabus
Even the most elegant autonomy-supportive design fails if people do not feel safe to try, err, and ask for help. In a field study of 51 teams within a single manufacturing company, Edmondson introduced the construct of team psychological safety and found it was associated with learning behaviors such as speaking up and seeking feedback, while team efficacy was not, once safety was accounted for (Edmondson, 1999). Learning behavior, in turn, mediated the link between psychological safety and performance. The study is cross-sectional and limited to one firm, so causation should be inferred with care. Yet as a foundational result it established the mechanism leaders must care about: people learn together when it is not personally risky to do so.
Translate that into the workplace. If employees fear that admitting confusion will hurt their standing, they will not practice new skills in the open. If teams punish experiments that miss, they will not find better methods. Safety is not softness. It is the precondition for disciplined learning, and a manager’s first job is to build it.
Structure follows strategy
Self-directed education inside firms runs into a structural barrier that too few leaders name. Traditional hierarchies are optimized for reliable execution of known tasks in stable settings. They are less suited to messy, cross-functional problem solving or to the development of broad-based judgment. That is the message of a broad scholarly synthesis on self-managing organizations by Lee and Edmondson. They define these organizations as ones that decentralize authority in a radical, formal, and systematic way, removing the manager–subordinate relationship as the core unit of control (Lee and Edmondson, 2017). The literature they review suggests conventional hierarchies deliver dependability but also reinforce status differences, inhibit complex problem solving across boundaries, and can stunt people’s growth.
This is not a call to abolish all hierarchy overnight. It is a reminder that learning and control are design choices that trade off. If a company wants employees who direct their own learning toward business outcomes, it must grant room to choose methods, share authority for decisions, and make it legitimate to surface and solve problems that do not fit existing boxes. Otherwise, self-directed education reduces to a slogan fought by the very systems it threatens.
A practical operating system for self-directed learning
Executives do not need a blank sheet. They need to wire three conditions into the flow of work and governance, then measure them as seriously as revenue.
-
Make autonomy real, but bounded. Use team charters and role canvases that specify clear goals and guardrails, then let people choose methods. This aligns with the autonomy dimension that intervention designers now systematically test, rather than leaving autonomy to chance or to poster slogans (Wang et al., 2024).
-
Invest in competence with visible practice. Replace occasional courses with regular, in-team practice loops: short cycles of attempting, reviewing, and adjusting work. Managers should model the behavior by asking for feedback on their own deliverables. This is how you surface learning behavior in the open, which is the path Edmondson linked to better performance in her field study (Edmondson, 1999).
-
Build relatedness through peer commitments, not pizza. Create small peer groups that set learning commitments and review progress together. Relatedness is not about social events. It is about feeling a mutual obligation to improve, which is one of the three needs now being tested across interventions at scale (Wang et al., 2024).
-
Redesign authority where it blocks learning. Push routine decision rights to the edge and formalize cross-functional ownership of problems. That direction is consistent with the defining feature of self-managing organizations, which remove the manager–subordinate dyad as the default mechanism of control in favor of more distributed authority for complex work (Lee and Edmondson, 2017).
Then measure what matters. If your engagement data looks like the world’s, with only 20 percent of people energised by their jobs, assume that the barrier is systemic. Gallup’s own figures show that some organizations achieve much higher engagement, including among managers, which underlines that practice, not context, differentiates outcomes (Gallup, 2025). Track leading indicators of the three psychological needs and of psychological safety at the team level, and do not bury the results. Treat them as operational metrics alongside customer and financial ones. When safety and needs satisfaction dip, expect learning to stall and fix the system where it fails.
The payoff is not a happier classroom feel. It is a more adaptive company. With engagement depressed and the cost of disengagement measured in trillions, firms that design for self-directed education will spend less time policing attendance and more time compounding capability where it matters. The choice is design or drift. One builds a bench of people who can see and solve the next problem without waiting to be taught. The other buys another course and hopes for the best.
Sources
- Gallup (2025). State of the Global Workplace 2025. Reports global engagement at 20 percent, an estimated 10 trillion dollars in lost productivity, and a steep drop in manager engagement, framing the scale and design nature of the problem. https://www.gallup.com/workplace/659610/state-global-workplace.aspx
- Wang et al. (2024). SDT-based educational interventions: a meta-analysis, Learning and Motivation 87. Synthesizes 36 intervention studies with 11,976 participants across autonomy, competence, and relatedness, showing the field’s shift to testing full-spectrum conditions for intrinsic motivation. https://www.sciencedirect.com/science/article/abs/pii/S0023969024000572
- Edmondson (1999). Psychological Safety and Learning Behavior in Work Teams, Administrative Science Quarterly 44. In a field study of 51 teams in one manufacturing company, psychological safety was associated with learning behavior, which mediated its link to performance, underscoring safety as a prerequisite for learning. https://journals.sagepub.com/doi/10.2307/2666999
- Lee and Edmondson (2017). Self-managing organizations: Exploring the limits of less-hierarchical organizing, Research in Organizational Behavior 37. Defines self-managing organizations and synthesizes evidence that traditional hierarchies suit reliable execution but can inhibit complex problem solving and development. https://www.sciencedirect.com/science/article/abs/pii/S0191308517300059