practiceDelivery & agility

Continuous Delivery

The capability to keep software in a releasable state so changes can be deployed safely, quickly, and sustainably whenever the organization chooses.

Download Markdown (.md)

01What it means

Continuous delivery is a set of technical and organizational practices that keeps software ready for release. Teams integrate small changes frequently, automate build and deployment processes, test at multiple levels, manage environments and configuration reproducibly, and receive fast feedback. Deployment may be continuous without every change being immediately released to all users.[1][2][3]

One artifact, increasing confidence Promote a reproducible change through checks until it is releasable
  1. 01IntegrateSmall versioned change
  2. 02Build onceCreate deployable artifact
  3. 03VerifyAutomated checks and evidence
  4. 04PromoteProduction-like environments
  5. 05Release safelyObserve, roll back, or recover forward

Continuous delivery keeps release a safe option; it does not require exposing every successful deployment to all users immediately.

02Why it matters

Large, infrequent releases accumulate risk and delay learning. A reliable delivery pipeline makes changes routine, exposes defects closer to their source, and allows product teams to respond to evidence without waiting for a release project. Smaller batches also make failures easier to diagnose and reverse.

03When to use it

Pursue continuous delivery for software products that need dependable, frequent change, including regulated systems when controls can be automated and audited. Introduce it incrementally where architecture or manual testing creates constraints. The goal is not maximum deployment frequency but safe, on-demand capability appropriate to the product.

04How it works

  1. Keep code, configuration, infrastructure definitions, and deployment scripts under version control.
  2. Integrate small changes frequently and run fast automated checks on every change.
  3. Build one deployable artifact and promote it through production-like environments.
  4. Automate repeatable deployments, database changes, security controls, and evidence capture where practical.
  5. Design for observability, backward compatibility, rapid rollback, or forward recovery.
  6. Improve the pipeline using lead time, deployment outcomes, failure data, and team feedback.

05A practical example

Example

A subscription team replaces monthly release weekends with a pipeline that builds, tests, scans, and deploys the same artifact through environments. A new cancellation option ships behind a flag to employees, then a small customer cohort, while monitoring and rollback controls protect the service.

06Where it breaks

Automation can accelerate bad changes if tests, observability, or product judgment are weak. Legacy coupling and manual assurance may require long-term investment. Pipeline metrics can be gamed, and continuous delivery does not mean bypassing accessibility, security, privacy, or change-management responsibilities.

Common mistakes

  • Equating continuous delivery with a CI server or deployment tool.
  • Automating a fragile process without simplifying its architecture and handoffs.
  • Using different artifacts or deployment methods in test and production.
  • Optimizing deployment count while ignoring reliability and customer outcomes.

07Sources

  1. What Is Continuous Delivery?Continuous Delivery · Authoritative practice overviewWhat Is Continuous Delivery?
  2. Principles of Continuous DeliveryContinuous Delivery · Authoritative principlesPrinciples of Continuous Delivery
  3. Continuous DeliveryDORA · Research-based capability guideContinuous Delivery
View saved topics

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