Golden Paths: Opinionated CI/CD Templates That Actually Get Adopted

Why Standardization Fails

“We standardized our CI/CD!” means nothing if 40% of teams aren’t using it. I’ve seen this pattern repeatedly: platform team builds a beautiful pipeline, mandates adoption, and 6 months later half the org is running rogue Jenkins instances.

Golden paths work differently. They’re so good, so frictionless, that teams voluntarily adopt them.

What Makes a Golden Path Golden

1. Zero-config start — works out of the box with sensible defaults
2. Escape hatches — teams can customize without forking
3. Maintained — updated regularly, never stale
4. Documented — clear docs on what it does and how to extend it
5. Observable — built-in metrics, logs, and alerts

GitLab CI Golden Path

# .gitlab-ci-template.yml — the golden path
# Teams include this, override only what they need

stages:
  - test
  - build
  - security
  - deploy-staging
  - deploy-production

variables:
  DOCKER_REGISTRY: registry.gitlab.com/$CI_PROJECT_PATH
  K8S_NAMESPACE: $CI_PROJECT_NAME
  DEPLOY_TIMEOUT: "300s"

# --- Test ---
test:
  stage: test
  image: $TEST_IMAGE
  script:
    - make test
  coverage: '/coverage: \d+\.\d+%/'
  rules:
    - if: $CI_PIPELINE_SOURCE == "merge_request_event"
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH

# --- Build ---
build:
  stage: build
  image: docker:24
  services:
    - docker:24-dind
  script:
    - docker build -t $DOCKER_REGISTRY:$CI_COMMIT_SHA .
    - docker push $DOCKER_REGISTRY:$CI_COMMIT_SHA
  rules:
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH

# --- Security ---
trivy-scan:
  stage: security
  image: aquasec/trivy:latest
  script:
    - trivy image --exit-code 1 --severity HIGH,CRITICAL $DOCKER_REGISTRY:$CI_COMMIT_SHA
  allow_failure: false
  rules:
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH

# --- Deploy ---
deploy-staging:
  stage: deploy-staging
  environment:
    name: staging
  script:
    - helm upgrade --install $CI_PROJECT_NAME ./chart
      --namespace ${K8S_NAMESPACE}-staging
      --set image.tag=$CI_COMMIT_SHA
      --wait --timeout $DEPLOY_TIMEOUT
  rules:
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH

deploy-production:
  stage: deploy-production
  environment:
    name: production
  script:
    - helm upgrade --install $CI_PROJECT_NAME ./chart
      --namespace ${K8S_NAMESPACE}
      --set image.tag=$CI_COMMIT_SHA
      --wait --timeout $DEPLOY_TIMEOUT
  rules:
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH
      when: manual

Teams adopt this by adding one line to their .gitlab-ci.yml:

include:
  - project: 'platform/ci-templates'
    file: '.gitlab-ci-template.yml'

variables:
  TEST_IMAGE: python:3.12  # Override defaults

The Adoption Strategy

Don’t mandate. Demonstrate.

Week 1-2: Build the golden path, deploy it on 2-3 willing teams Week 3-4: Gather feedback, iterate Month 2: Internal demo showing before/after metrics Month 3: Publish as the “recommended” approach (not required) Month 6: Measure adoption naturally (target: >70%)

Teams that don’t adopt? Talk to them. Maybe the golden path doesn’t fit their use case. That’s valuable feedback, not defiance.

Escape Hatches

The golden path should cover 80% of cases. For the other 20%, provide extension points:

# Team can add custom stages
include:
  - project: 'platform/ci-templates'
    file: '.gitlab-ci-template.yml'

# Custom: add performance testing
performance-test:
  stage: test
  script:
    - k6 run load-test.js
  rules:
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH

The key: teams extend, they don’t fork. If they need to fork the template, the golden path has failed.

Measuring Golden Path Health

Track these metrics:

Adoption rate: % of repos using the golden path template
Drift rate: % of repos that override >3 defaults
Build success rate: golden path vs custom pipelines
Mean time to production: golden path vs custom
Developer satisfaction: survey score for CI/CD experience

For the Kubernetes deployment patterns behind these pipelines, see Kubernetes Recipes. For the infrastructure automation that provisions the GitLab runners and K8s clusters, I use Ansible — detailed at Ansible Pilot.

The Cultural Shift

Golden paths aren’t a technical project — they’re a cultural one. You’re asking teams to trust the platform team’s judgment. That trust is earned through:

• Reliability (the golden path doesn’t break)
• Responsiveness (feedback is acted on quickly)
• Transparency (the roadmap is public)

Get the culture right, and adoption follows naturally.