Q1: Dominant flow constraint in split-authority delivery systems

Research Question

What is the dominant source of delay and instability in split-authority delivery systems: capacity shortage, dependency coupling, approval latency, funding gates, or fragmented decision rights?

A split-authority delivery system is one in which no single actor owns risk, cost, and execution simultaneously, meaning authority is divided among at least two stakeholder groups, each with independent veto or approval power, typically a business owner, a technology delivery team, and an independent risk or compliance function.

Findings

Executive Summary

In split-authority delivery systems, governance-generated queueing caused by approval latency and fragmented decision rights is the dominant throughput constraint, not capacity shortage. DORA's multi-firm research confirms that external change approvals are negatively correlated with all delivery performance metrics, and a value stream mapping (VSM) case study of a 200-engineer component-team organisation found 97% wait time, consistent with the governance-constraint direction of the DORA evidence. Fragmented decision rights are the organisational root cause of approval latency: when no single actor owns risk, cost, and execution simultaneously, every substantive decision must travel a multi-party approval circuit whose cumulative wait time dominates total cycle time. Dependency coupling amplifies the effect by increasing the number of approval nodes each item crosses, and funding gates contribute periodic batch-demand instability as a coarser-cycle variant of the same mechanism. Capacity shortage is secondary and partially endogenous: governance overhead consumes productive capacity, and the approval queue obscures latent capacity that would otherwise be measurable.

Key Findings

1. Approval latency is a policy constraint in the Theory of Constraints (TOC) sense, meaning it is rule-imposed rather than capacity-limited, and therefore blocks throughput even when physical execution capacity is sufficient.

2. DORA multi-year, multi-firm survey research finds that external change approval processes are negatively correlated with all four software delivery performance metrics, confirming that governance-type constraints suppress measured delivery performance across diverse organisations.

3. Value stream mapping (VSM) analysis applied to a 200-engineer organisation with tightly coupled component teams found 97% waste or wait time, meaning months of calendar time for work requiring only 2 to 3 days of active effort, demonstrating the system-level magnitude of governance-generated queue latency.

4. Fragmented decision rights are the organisational root cause of approval latency in split-authority systems: when no party holds sufficient authority to commit unilaterally, every substantive decision creates a multi-party approval circuit that adds queue wait time at each step.

5. Dependency coupling amplifies governance-generated queueing by adding one or more cross-team synchronisation queues to every work item that crosses a team boundary, compounding the total wait time produced by the governance approval chain.

6. Funding gates create throughput instability through a batch demand spike mechanism: demand accumulates during gate closure and enters the delivery system simultaneously at gate opening, overwhelming execution capacity even in organisations with adequate steady-state bandwidth.

7. Capacity shortage is secondary and partially endogenous in split-authority systems: governance overhead consumes productive delivery capacity and the approval queue conceals latent capacity, so the constraint appears to be execution bandwidth when the binding limiter is actually governance-generated blocking.

8. TOC warns that adding capacity before identifying and exploiting the actual constraint is the most common improvement failure, and organisations that add headcount without reducing governance overhead typically find that throughput improves less than expected because the governance constraint absorbs the new capacity.

9. The feedback loop between approval latency and large-batch delivery is self-reinforcing: high approval latency incentivises teams to bundle work into fewer, larger submissions; larger batches increase change failure risk; higher failure risk makes approvers more conservative; and conservative approvers extend cycle times further.

10. There is a non-zero minimum approval-latency floor in regulated enterprises set by mandatory controls such as segregation of duties and change control for regulated systems, but most split-authority organisations operate substantially above this floor because governance patterns evolved to apply heavyweight controls uniformly rather than proportionately to transaction risk.

Assumptions

• The five candidate constraint categories (capacity shortage, dependency coupling, approval latency, funding gates, fragmented decision rights) are collectively exhaustive for the scope of this analysis; organisations whose primary instability source is technology failure or market-demand collapse are out of scope.
• DORA survey research is a reasonable proxy for the typical split-authority organisation because high governance overhead correlates with lower performance in the dataset and the sample spans multiple industries and firm sizes.

Analysis

The five candidate constraints form a hierarchy rather than a set of independent alternatives: fragmented decision rights are the structural condition; approval latency is the direct mechanism; dependency coupling is the amplifier; and funding gates are the periodic, coarser-cycle variant. Capacity shortage is the only independent candidate, and the evidence does not support it as the primary system-level constraint in split-authority organisations that have not yet reduced governance friction.

DORA's negative correlation finding is the most robust evidence source in this item because it is multi-firm, multi-industry, and longitudinal, and because it directly tests the governance-approval variable rather than inferring it. TOC policy-constraint theory and the VSM case study evidence are mechanistically consistent with the DORA finding: all three identify wait time in governance queues as the dominant cycle-time contributor. Prior completed repository items on governance controls effectiveness and governance failure mechanisms provide additional causal texture: controls become overhead when applied uniformly at high volume, which is the defining condition in split-authority systems where every work item passes through a multi-stakeholder approval chain.

Capacity shortage does not have strong evidentiary support as the primary constraint in split-authority systems that have not yet reduced governance friction. TOC explicitly warns that adding capacity before exploiting the actual constraint is a common and costly misdiagnosis. The IT throughput constraint completed item notes that governance overhead consumes capacity and that the approval queue hides latent capacity, which is the mechanistic pathway by which a capacity-shortage appearance masks a governance-constraint reality.

The self-reinforcing feedback loop between approval latency and batch-size growth is the instability mechanism: any increase in governance overhead accelerates the feedback, driving throughput down and batch size up, until an external intervention (governance redesign, senior escalation, or accumulated delivery failure) breaks the cycle.

Risks, Gaps, and Uncertainties

• No primary cross-firm experimental study isolates approval latency as the binding constraint in split-authority organisations through direct manipulation. All evidence is observational, case-based, or theoretical.
• Precise magnitude decomposition (what fraction of cycle time each constraint category accounts for) is not supported by the available evidence base.
• The secondary status of capacity shortage is directional, not universal: in severely understaffed teams, capacity may be independently binding, and the findings here apply primarily to organisations that have not yet addressed governance friction.
• The feedback loop mechanism between approval latency and batch-size growth is well-supported mechanistically but has not been experimentally isolated from other explanatory variables in a controlled study.

Open Questions

• At what threshold of governance overhead does capacity shortage become co-dominant with approval latency, and is this threshold measurable using available metrics such as DORA lead time or WIP levels?
• How does the constraint hierarchy differ in heavily regulated sectors (banking, healthcare, public sector) versus less-regulated technology firms, given that the mandatory approval floor is higher in the former?
• Does the introduction of AI-assisted delivery tooling shift the dominant constraint from approval latency toward capacity shortage as governance becomes the clearer residual bottleneck?

Output

knowledge. This item identifies the dominant flow constraint in split-authority delivery systems and establishes the causal hierarchy among the five candidate categories, providing the empirical and theoretical foundation for demand segmentation (Q2), exception routing design (Q3), decision rights placement (Q4), and leading indicator selection (Q6).

sources

• [x] DORA (2019) Streamlining Change Approval - canonical DevOps Research and Assessment (DORA) guidance on change approval processes and their effect on software delivery performance.
• [x] DORA (2019) State of DevOps Report - survey-based report finding that external change approval is negatively correlated with software delivery performance.
• [x] DORA Loosely Coupled Teams - DORA capability page on architectural and organisational conditions associated with high delivery performance.
• [x] DORA Metrics: The Four Keys - DORA definition of throughput and instability metrics for software delivery.
• [x] Theory of Constraints Institute: Five Focusing Steps - canonical description of the Theory of Constraints Process of Ongoing Improvement (POOGI).
• [x] Org Topologies case study: component teams to Team Topologies - case study reporting 97% waste or wait time under a tightly coupled component-team architecture.
• [x] Mitchell (2026) Backpressure Infrastructure and the Theory of Constraints - prior completed repository item on TOC mechanisms, Drum-Buffer-Rope scheduling, and policy versus physical constraints.
• [x] Mitchell (2026) Operating model synthesis for split-authority delivery systems - prior completed synthesis item identifying governance-generated queueing as the dominant constraint.
• [x] Mitchell (2026) Governance controls effectiveness conditions - prior completed item on when controls minimise versus amplify coordination cost.
• [x] Mitchell (2026) Governance failure mechanisms and bureaucracy circumvention - prior completed item on how controls become overhead and staff circumvent them.
• [x] Mitchell (2026) Funding authority and delivery-risk accountability split - prior completed item on the delivery consequences of separating funding authority from delivery accountability.
• [x] Mitchell (2026) IT throughput constraint magnitude and debt accumulation rate - prior completed item on throughput constraints and their relationship to shadow IT demand.
• [x] Mitchell (2026) Organisational failure modes: customer-segment demand versus domain-based IT teams - prior completed item on boundary-mismatch delivery bottlenecks.