One of the more persistent misreadings of agile is that it is anti-manager. Self-organizing teams, the story goes, make bosses redundant. The board runs the work, the team runs itself, and the manager becomes a relic to be managed out in the next reorganization. Three rounds of tech layoffs have given that story a grim second life, with delayering pitched as agility and agility used to justify delayering.
The research does not support the fantasy. It supports something more demanding. Agile does not remove leadership from the system. It changes the kind of leadership the system rewards, and it punishes the old kind more visibly than any org chart ever did.
Instability is a leadership failure, and it is expensive
Start with what breaks agile teams in practice. The DORA State of DevOps report (2024), drawing on roughly 39,000 respondents, found that instability in organizational priorities substantially reduces productivity and increases burnout. The finding that should stop a leadership team cold is what came next. That damage persisted even when teams had good leadership and good documentation. No mitigating factor rescued a team whose priorities kept changing underneath it.
Read carefully, this is a report about management even though it is framed as a report about engineering. Priorities do not become unstable by accident. They wobble because leaders will not choose, because too many stakeholders can inject work, because saying no upward is harder than saying yes and passing the churn down. The ceremony cannot fix that. No length of standup stabilizes a roadmap that leadership refuses to hold still. This is squarely a job for the people with the authority to decide what the organization is not going to do this quarter, and to protect that decision.
It is worth being precise about the strength of this evidence. The DORA data is a cross-sectional survey, so it establishes that instability and poor outcomes travel together rather than proving one causes the other in every case. But the size of the sample and the consistency of the pattern make it hard to wave away, and the direction of the relationship matches what the scaling literature reports independently. When a finding this large points at leadership, the burden shifts to anyone who wants to keep blaming the team.
The leadership that helps is not the leadership that controls
So leadership matters. The question is which kind. Here the same DORA report (2024) is specific. Transformational leadership, meaning leaders who inspire, intellectually stimulate their teams and offer genuine support, was associated with significantly higher productivity, better organizational performance and higher satisfaction, along with lower burnout. The human factor sat above process and tooling in explaining delivery outcomes.
This is the manager's job rewritten in a single sentence. The role is not to assign tasks and check attendance. It is to set a stable direction, remove obstacles, develop people, and then get out of the way of a team that has been given the authority to do the work. That is harder than command, not easier, which is why so many organizations quietly keep the command and merely rename it.
Support is a success factor, not a nicety
The scaling literature reaches the same place from a different door. Dikert, Paasivaara and Lassenius (2016), in their review of 42 industrial cases of large-scale agile adoption, list management support as one of the recurring success factors, alongside customization, coaching and a real shift in mindset. The transformations that worked were not the ones with the most disciplined ceremonies. They were the ones where leadership visibly backed the change, invested in it, and adapted it to the context instead of imposing a framework and walking away.
What kind of support? Steegh, Van De Voorde and Paauwe (2025) sharpen the answer in their review of how agile teams reach effectiveness, where shared leadership emerges as a core enabler. Shared leadership does not mean leaderless. It means leadership that is distributed and enabled rather than concentrated and imposed, which requires a formal manager who is secure enough to let authority flow to the team and skilled enough to know when to step back in. Abdication is not delegation. A manager who disappears has not enabled shared leadership, only removed the support the DORA data says teams need.
The practical read
Put these together and the modern manager's job description writes itself, though few will enjoy reading it. Hold priorities stable, because the DORA report (2024) shows that instability alone can sink a team no matter how good everything else is. Lead by inspiration and support rather than surveillance, because that is the leadership style the same evidence rewards. Back the change with resources and adaptation, because that is what separated the successful cases in Dikert, Paasivaara and Lassenius (2016). And distribute authority without abandoning the team, because shared leadership, in Steegh, Van De Voorde and Paauwe (2025), is enabled, not merely announced.
The manager who fears agile has usually mistaken the job for control. The manager agile actually needs does something rarer and more valuable. They make it possible for a team to lead itself, and they hold the line on priorities while it does.
Sources
- DORA / Google Cloud (2024). Accelerate State of DevOps Report 2024. Survey of roughly 39,000 respondents finding priority instability harms productivity and burnout even under good leadership, while transformational leadership is associated with better performance and lower burnout. https://dora.dev/research/2024/dora-report/
- Dikert, Paasivaara and Lassenius (2016). Challenges and success factors for large-scale agile transformations: A systematic literature review. Review of 42 industrial cases identifying management support and customization among the recurring success factors. https://doi.org/10.1016/j.jss.2016.06.013
- Steegh, Van De Voorde and Paauwe (2025). Understanding how agile teams reach effectiveness: A systematic literature review to take stock and look forward. Identifies shared leadership as a core enabler of agile team effectiveness, meaning distributed rather than absent leadership. https://doi.org/10.1016/j.hrmr.2024.101056