Executives are reimposing office days and tightening budgets, yet product teams are still told to "do agile" as if a cadence of ceremonies could neutralize volatility. It cannot. When priorities ricochet week to week, no framework turns churn into progress. Large-scale evidence now says the quiet part out loud: stability and leadership quality outweigh ritual.
The ritual trap
The scaling playbooks look polished. The research behind them is not. A systematic literature review of large-scale agile transformations found that nearly 90 percent of the underlying publications were practitioner reports rather than academic studies, a weak base for sweeping prescriptions (Dikert, Paasivaara and Lassenius, 2016). Even so, across 42 industrial cases the themes repeat with consistency: transformations stall without visible management support, careful customization of the model, training and coaching, and a shift in mindset and alignment across functions (Dikert, Paasivaara and Lassenius, 2016). In other words, ceremonies fail when the surrounding system stays the same.
The common pattern is familiar. Boards demand speed. Leaders buy a framework. Teams adopt the rituals. Yet dependencies remain, incentives clash, and priorities change mid-sprint. Blaming the standup is tidy. It is also wrong on the evidence.
What the strongest data is telling you
The largest contemporary dataset on software delivery and operations points to a more prosaic root cause. In a survey of roughly 39,000 respondents, instability of organizational priorities was associated with lower productivity and higher burnout, and this negative relationship persisted even where respondents reported strong leadership and good documentation (DORA, 2024). The signal is uncomfortable. You cannot schedule your way out of strategic whiplash. If the work people are asked to do keeps changing, their energy and output will fall.
Leadership still matters, just not as a fig leaf for chaos. The same research associates transformational leadership, the kind that inspires, challenges, and supports teams, with higher productivity, stronger organizational performance, and greater satisfaction, alongside lower burnout (DORA, 2024). This is not a plea for charismatic speeches. It is a reminder that the human system around the process is a performance variable in its own right.
There is a second pillar to take seriously. Before most companies had even coined their scaling frameworks, multi-year analyses of delivery teams argued that team performance can be measured and improved through specific organizational capabilities, and that doing so contributes to competitive advantage (Forsgren, Humble and Kim, 2018). The authors base their case on four years of research using State of DevOps data from 2014 to 2017 and statistical methods, and they explicitly claim to identify measurable team performance metrics and capabilities that causally drive higher software delivery performance (Forsgren, Humble and Kim, 2018). The causal language is the authors’ claim rather than a settled consensus, but the implication for leaders is clear enough. If you cannot describe and track how work flows in your environment, agile becomes theater.
Fit beats fidelity
The large-scale transformations reviewed in industry point to a simple design rule: tailor before you scale. Successful cases customized their models to context, invested in coaching, and worked on mindset and alignment, not just mechanics (Dikert, Paasivaara and Lassenius, 2016). That is a harder ask than rolling out identical boards and ceremonies. It forces choices about how funding, architecture, and decision rights will really work.
Start with stability. Given the association between shifting priorities and worse outcomes, set a hard constraint on churn. Define what “frozen” means for your planning horizon and hold to it except under clear escalation. Make stability a management responsibility rather than a team burden. Treat every unplanned pivot as a leadership debt to be repaid with explicit de-prioritization elsewhere. These are managerial commitments, not backlog hygiene.
Then look at leadership practice. The patterns linked with better outcomes in the recent evidence are specific: leaders who set purpose, stimulate teams intellectually, and provide support create better conditions for delivery and reduce burnout (DORA, 2024). That shows up in the day to day as clarity on outcomes, room for teams to propose technical approaches, and help removing organizational blockers. It is not a transformation office memo. It is behavior.
Finally, measure what you intend to improve, and make those measures fit your system. The authors behind Accelerate argue that delivery performance rests on measurable team metrics and organizational capabilities, built from multi-year analysis of industry data, and they presented their program strongly enough to win recognition such as the Shingo Institute Publication Award (Forsgren, Humble and Kim, 2018). Treat this as a hypothesis to operationalize in your context. Decide which few measures make sense for your architecture and release model, instrument them, and inspect them with the same regularity you give to ceremonies.
A practical contract for real agility
If you want agile beyond the ritual, write down a compact between executives, managers, and teams that reflects what the evidence suggests.
- Executives commit to priority stability over a defined horizon, with a visible process for trade-offs. This addresses the strongest drag on productivity and well-being observed in large-scale survey data (DORA, 2024).
- Managers practice transformational leadership in concrete ways: outcome clarity, intellectual challenge, and support. This correlates with better productivity, performance, and satisfaction, and with lower burnout (DORA, 2024).
- Teams get customization over strict fidelity. Adapt planning and roles to your domain, and fund coaching so the new habits stick, consistent with success factors in large-scale transformations (Dikert, Paasivaara and Lassenius, 2016).
- Everyone agrees to measure flow and outcomes, not ceremony completion. Use a small set of metrics and capabilities appropriate to your system, drawing on the claims and framework advanced by multi-year DevOps research, then test and refine locally (Forsgren, Humble and Kim, 2018).
This is not romantic. It is operational. The first three bullets reduce noise, the last one raises the signal. If your standups are humming and your output still feels brittle, stop tuning the ritual. Stabilize what matters, show up as a leader, tailor the model, and measure what you are actually trying to improve. That is the agile worth keeping.
Sources
- Dikert, Paasivaara and Lassenius (2016). Challenges and success factors for large-scale agile transformations, Journal of Systems and Software. A review of 52 publications covering 42 cases finds a weak evidence base dominated by practitioner reports and highlights management support, customization, coaching, and mindset/alignment as recurring success factors and challenges. https://www.sciencedirect.com/science/article/pii/S0164121216300826
- DORA / Google Cloud (2024). Accelerate State of DevOps Report 2024. A large cross-sectional survey (~39,000 respondents) associates unstable priorities with lower productivity and higher burnout, and links transformational leadership with better productivity, performance, and satisfaction, and lower burnout. https://dora.dev/research/2024/dora-report/
- Forsgren, Humble and Kim (2018). Accelerate: The Science of Lean Software and DevOps. The book presents four years of research using State of DevOps data, claims to identify measurable metrics and organizational capabilities that causally drive higher delivery performance, and argues that delivery capability is a competitive advantage. https://itrevolution.com/product/accelerate/