practiceDelivery & agility

Dependency Management

The practice of identifying, reducing, sequencing, and monitoring relationships that constrain independent product delivery.

Download Markdown (.md)

01What it means

Dependency management addresses situations where an outcome or delivery requires work, information, approval, infrastructure, or capability controlled elsewhere. It combines visibility and coordination with redesign of teams, architecture, scope, or decisions so avoidable dependencies are reduced rather than merely tracked.[1][2]

Owned relationships around a blocked outcome Expose what is needed, who controls it, and how the dependency can be reduced
Outcome at risk
Upstream work
Required capability or deliverable
Decision or approval
Owner, lead time, acceptance
Infrastructure
Capacity, environment, interface
Delivery plan
Forecast and sequencing change
Risk review
Warning signals and escalation
Redesign options
Decouple, rescope, share skills

The goal is not a larger dependency log; it is earlier integration and removal of avoidable structural coupling.

02Why it matters

Hidden dependencies create waiting, integration failures, conflicting priorities, and unreliable forecasts. Making them explicit supports realistic sequencing and ownership, while reducing them improves flow, resilience, and teams’ ability to deliver customer value independently.

03When to use it

Use dependency management whenever work crosses teams, systems, vendors, governance bodies, or release trains. Apply it early during roadmapping and planning, then revisit it during discovery and delivery. Lightweight coordination is enough when the relationship is low-risk and easily reversible.

04How it works

  1. Identify what is needed, from whom, by when, and what outcome is blocked if it is unavailable.
  2. Classify dependencies by type, criticality, lead time, uncertainty, and whether they are genuinely unavoidable.
  3. Assign owners on both sides and agree interfaces, decision rights, acceptance conditions, and communication cadence.
  4. Reduce the dependency through scope changes, decoupled architecture, shared skills, enabling teams, or different sequencing where practical.
  5. Visualize remaining relationships on the roadmap or delivery plan and incorporate them into forecasts and risk reviews.
  6. Integrate early, monitor warning signals, and escalate conflicts while useful options still remain.
  7. After delivery, address recurring structural causes instead of normalizing coordination overhead.

05A practical example

Example

A mobile checkout improvement depends on identity, payments, legal review, and analytics. The team agrees stable interfaces with identity, tests the payment integration early, brings legal review into discovery, and removes a custom analytics dependency by using an existing event standard. Only the unavoidable release sequence remains on the plan.

06Where it breaks

Dependency registers can become administrative clutter or justify centralized control. Visualizing every relationship obscures the few that threaten outcomes, while excessive coordination consumes delivery capacity. Some dependencies reflect sensible shared capabilities, so eliminating all of them may duplicate systems and expertise.

Common mistakes

  • Recording dependencies without naming reciprocal owners and required decisions.
  • Discovering integration needs only near the release date.
  • Treating every dependency as fixed instead of redesigning avoidable ones.
  • Tracking low-impact links while ignoring priority conflicts and long lead times.

07Sources

  1. Project DependenciesAtlassian · Professional guideProject Dependencies
  2. Git and Project DependenciesAtlassian · Technical dependency guideGit and Project Dependencies
View saved topics

Start typing to search titles, definitions, categories, and frameworks.