Executives keep asking why sprint rituals have not delivered speed while AI bets pile up, return-to-office mandates soak up attention, and hiring is tight. The standups happen. The backlog still bloats. The release train is late again. The problem is not the cadence. It is the system the rituals are trying to paper over.
The ritual trap is a scaling mirage
If agile-at-scale were solved by choreography, the literature would say so. It does not. A systematic review of large-scale agile transformations that synthesized 52 publications across 42 industrial cases found the evidence base is thin, with almost 90 percent of included papers being practitioner reports rather than academic studies (Dikert, Paasivaara and Lassenius, 2016). That weakness matters because it tempts leaders to copy ceremonies instead of changing conditions. Even so, the review isolates what actually shows up when change sticks: visible management support, customization of the model to context, training and coaching, and a mindset shift with alignment across functions (Dikert, Paasivaara and Lassenius, 2016). In other words, the levers are organizational, not theatrical.
This is why many scaling efforts stall. You can impose synchronized standups and quarterly planning and still have teams blocked by architecture, conflicting incentives, and drive-by reprioritizations. Rituals cannot cancel dependencies. Only structure, leadership, and the technical pathways of work can do that.
What the 2024 data says about the real constraints
Fresh large-sample data is blunt about the harm done by context. The 2024 State of DevOps report finds that unstable organizational priorities correlate with lower productivity and higher burnout, and that this effect persists even with strong leadership and good documentation in place (DORA, 2024). That persistence matters for every agile transformation. It says you cannot compensate for priority churn by turning the cadence dial or writing more playbooks.
The same dataset links transformational leadership to better delivery productivity, stronger organizational performance, higher satisfaction, and lower burnout (DORA, 2024). The behaviors that help are not ceremony policing. They are leaders who set a compelling direction, stimulate teams intellectually, and provide support. Again, the lever is human and organizational.
There is a third signal in the findings that is often misread. Internal developer platforms are associated with higher productivity at the individual, team, and organizational levels, yet they may introduce trade-offs in stability or throughput that leaders need to manage carefully (DORA, 2024). The point is not to chase a platform for its own sake. The point is to build self-service paths that remove waiting, reduce cognitive load, and make good defaults easy, while watching the operational consequences.
All of these relationships are correlational, not causal. The report is explicit about its survey design. Treat them as conditions to engineer toward and tensions to balance, not as a recipe that guarantees a specific outcome in your firm (DORA, 2024).
Architecture over choreography
The case for shifting attention from ceremonies to capabilities is not new. Over four years of research analyzed in Accelerate argues that software delivery performance can be measured reliably and that specific organizational capabilities drive it, challenging the old belief that delivery team performance does not confer competitive advantage (Forsgren, Humble and Kim, 2018). The book’s core claim is causal, though readers should note it is the authors’ position based on their analysis. Its managerial implication is clear enough: stop judging agile by ritual compliance. Judge it by delivery performance, and build the capabilities that move those metrics.
Taken together, the signals point in the same direction. If you want agile beyond the ritual, you must shrink dependencies, stabilize intent, and give teams paved roads. That is architecture and leadership, not more ceremonies.
A practical playbook to scale autonomy
-
Stabilize intent on a clear tempo. The single strongest drag in the 2024 data is unstable priorities, which correlate with lower productivity and more burnout, even when leadership and documentation are strong (DORA, 2024). Set explicit windows in which portfolio priorities are fixed, with pre-agreed exceptions. Publish changes as product decisions, not hallway requests.
-
Build a platform as a product, not a tool shelf. Treat internal developer platforms as customer-facing products with service levels, product management, and user research. Their value is in self-service: templates, golden paths, and automated guardrails that make the right thing the fast thing. Pay attention to the trade-offs. Platforms can lift productivity, but leaders must monitor stability and throughput and adjust defaults, limits, and education accordingly (DORA, 2024).
-
Lead like the work is complex, not controllable. Transformational leadership correlates with better delivery outcomes and lower burnout (DORA, 2024). That means removing obstacles, asking for learning over blame, and coaching teams on intent and constraints rather than prescribing tasks. In the large-scale review, management support, coaching, and a mindset shift toward alignment were all common success factors across industrial cases, which reinforces that leadership behavior is not a soft add-on. It is central to agility at scale (Dikert, Paasivaara and Lassenius, 2016).
-
Customize the framework to your system. The systematic review flags customization and coaching as recurrent success factors and highlights how often off-the-shelf adoptions struggle (Dikert, Paasivaara and Lassenius, 2016). Start with your real constraints: where work waits, who must sign off, what depends on what. Then tune roles, cadences, and artifacts to fit those specifics. Copying ceremonies without changing the system is theatrics.
-
Measure delivery, not attendance. Use a small, stable set of delivery performance measures, and tie improvement efforts to capabilities that research argues matter for throughput and reliability, not to the volume of meetings attended. Accelerate’s central contention is that these capabilities can drive performance in ways that matter competitively, which is the point of agility in the first place (Forsgren, Humble and Kim, 2018).
The through line is simple. Agile rituals are cheap to schedule and comforting to audit. They are also the least powerful lever you have. The stronger levers are stabilizing priorities, building self-service platforms that remove toil and waiting, and leading in ways that create clarity and support. The literature on large-scale transformations, limited as it is, pushes you there. The 2024 data on what teams experience pushes you there too. If you want agility that survives board pressure, AI deadlines, and the next reorg, scale autonomy with platforms and leadership. The ceremonies can follow, not lead.
Sources
- Dikert, Paasivaara and Lassenius (2016). Challenges and success factors for large-scale agile transformations, Journal of Systems and Software. Synthesizes 42 industrial cases, finds the evidence base weak and highlights management support, customization, coaching, and alignment as recurring success factors, supporting the argument to fix context over rituals. https://www.sciencedirect.com/science/article/pii/S0164121216300826
- DORA (2024). Accelerate State of DevOps Report 2024. Reports that unstable priorities correlate with lower productivity and higher burnout even with strong leadership and documentation, that transformational leadership correlates with better outcomes, and that internal platforms lift productivity with trade-offs to manage, grounding the piece’s 2020s frame. https://dora.dev/research/2024/dora-report/
- Forsgren, Humble and Kim (2018). Accelerate: The Science of Lean Software and DevOps. Argues that delivery performance is measurable and that organizational capabilities can drive it, challenging ritual compliance as the yardstick and supporting a focus on capabilities and outcomes. https://itrevolution.com/product/accelerate/