Return-to-office mandates and morale campaigns have not stopped the slide. Global employee engagement fell to 20 percent in 2025, the lowest since 2020, with Gallup estimating about 10 trillion dollars in lost productivity, roughly 9 percent of world GDP (Gallup, 2025). Managers themselves are running out of steam. Their engagement dropped from 27 to 22 percent between 2024 and 2025, even as best-practice organizations report 79 percent manager engagement, a gap that points to design, not destiny (Gallup, 2025). When attention fixes on attendance or slogans, the real constraint goes untouched. The bottleneck is where decisions live.
Hierarchies execute well, then stall on complexity
Conventional hierarchies were built for reliable execution of known tasks in stable environments. They do that well. They struggle when the work is cross-functional, uncertain, and changing. That is not ideology, it is the pattern in the research synthesis on self-managing organizations, which defines these firms by the radical, formal, and systematic decentralization of authority and the elimination of the manager–subordinate relationship as a structural default (Lee and Edmondson, 2017). The authors summarize decades of management studies: managerial hierarchies tend toward rigidity and status differences that dampen development, and they inhibit complex problem solving that cuts across functions (Lee and Edmondson, 2017).
If you ask managers to be the approvals desk for everything, you create queueing, learned helplessness, and brittle coordination. If you move decisions into the work with explicit mechanisms, you shorten loops and surface talent. That is the through line from self-management research to the engagement data. Gallup’s figures are descriptive, not causal, but the contrast they report between average practice and best practice is consistent with a conclusion that design choices, not vibes, drive whether people feel their effort matters (Gallup, 2025).
Let people shape their jobs, then watch engagement carry performance
Empowerment is easiest to critique when it is vague. It gets harder to dismiss when you define it as a set of teachable behaviors. Job crafting is one such set, and we now have a meta-analysis of interventions that teach people to reshape task boundaries, relationships, or perceptions of their roles. Across 14 intervention studies, average effects are small but real: job crafting behaviors increase by g = 0.26, work engagement by g = 0.31, and contextual performance by g = 0.39. The lower bound of the confidence interval for contextual performance is 0.01, so treat that estimate with caution, and the authors show that performance gains are fully mediated by engagement rather than by job crafting behaviors on their own (Oprea, Barzin, Vîrgă, Iliescu and Rusu, 2019). Their utility analysis suggests meaningful economic returns for healthcare organizations, even at these small effect sizes (Oprea et al., 2019).
Two implications follow. First, you do not need to reorganize the firm to start moving power into the work. Teaching people to reshape their roles within clear constraints can lift engagement and, through that channel, performance. Second, the mechanism matters. Crafting appears to work by increasing energy and involvement, not by magic. If engagement is the engine, then any structure that deadens it will blunt your performance.
Safety is the scaffold that makes empowerment real
Authority without safety is theater. Teams will not take ownership if speaking up or experimenting carries social risk. The construct of psychological safety was named and measured in a field study of 51 teams in a single manufacturing company. The results are clear about mechanism, not causation across all settings. Teams with higher psychological safety showed more learning behavior, and that learning behavior mediated the relationship with performance, while collective confidence alone did not predict learning once safety was accounted for (Edmondson, 1999). This is a classic result for a reason. It explains why self-management and job crafting succeed in some pockets and stall in others. If people expect interpersonal risk, they withhold questions, hide errors, and stop adapting. If they expect candor will not be punished, they surface problems early and coordinate.
Connect these dots and an operating picture emerges. Self-managing designs relocate authority away from a single hierarchical node and toward roles and processes that sit with the work. Job crafting gives individuals a method to tune those roles to reality. Psychological safety supplies the social conditions under which teams will actually use that authority and method to learn. None of this relies on charisma. All of it can be designed.
How to move power into the work, now
Given the slump in engagement and the strain on managers, act like an engineer, not a therapist. Start small, measure, and scale what travels.
- Define decisions, then decentralize them: Map your highest-friction approvals. For each, specify who decides, on what criteria, and with what information. Where feasible, move decision rights to the roles that create or consume the information, a practice aligned with the formal decentralization that distinguishes self-managing organizations (Lee and Edmondson, 2017). Publish the map.
- Run job crafting sprints: Offer structured job crafting workshops to targeted teams. Expect modest average gains, because that is what the meta-analysis found, and track engagement as the lead indicator since it mediates performance effects (Oprea et al., 2019). In healthcare and similar settings the returns can be economically significant; in other contexts, test and verify.
- Build psychological safety into routines: Do not preach it, stage it. Use lightweight practices that raise the odds of learning behavior, such as explicit check-ins for risks and near-misses, joint problem-solving reviews, and norms that reward question-asking. The goal is to increase the behaviors that mediate performance, consistent with the original field evidence (Edmondson, 1999).
- Watch the manager role shrink and sharpen: As decision rights move and teams craft jobs, managers should carry fewer approvals and more coaching on priorities and trade-offs. The Gallup data on collapsing manager engagement are descriptive, yet they point to a design failure. In firms that redesign the role, engagement among managers rises toward best-practice levels, a reminder that structure can enable or exhaust managers (Gallup, 2025).
You cannot survey your way out of a power problem. The global cost of disengagement is too large to ignore, but the fix is not another campaign. It is a shift in who signs off, how jobs flex, and whether teams feel safe enough to learn in public. Move authority into the work, equip people to shape that work, and make it safe to try. Engagement will follow, and with it, the performance you actually care about.
Sources
- Gallup (2025). State of the Global Workplace 2025. Reports global engagement at 20 percent, a 10 trillion dollar productivity cost, and a sharp drop in manager engagement, framing the design problem this piece addresses. https://www.gallup.com/workplace/659610/state-global-workplace.aspx
- Lee and Edmondson (2017). Self-managing organizations: Exploring the limits of less-hierarchical organizing. Defines SMOs as formally decentralizing authority and summarizes why managerial hierarchies struggle with complex, cross-functional work. https://www.sciencedirect.com/science/article/abs/pii/S0191308517300059
- Oprea, Barzin, Vîrgă, Iliescu and Rusu (2019). Effectiveness of job crafting interventions: a meta-analysis and utility analysis. Finds small but significant gains in job crafting, engagement, and contextual performance, with performance mediated by engagement and economic value in healthcare contexts. https://doi.org/10.1080/1359432X.2019.1646728
- Edmondson (1999). Psychological Safety and Learning Behavior in Work Teams. Shows that team psychological safety predicts learning behavior, which mediates performance, clarifying the social conditions under which empowerment takes hold. https://journals.sagepub.com/doi/10.2307/2666999