Back to Audits

Deploy & Observe

Can we run and monitor this?

Review deployment pipelines, monitoring coverage, and resilience patterns. Ship confidently and know when things break.

Stage expectations

POC

Skip check

  • Manual deploy is acceptable

  • No monitoring required

  • Just get it running

MVP

Light check

  • Basic CI/CD in place

  • Error tracking active

  • Can deploy without fear

MMP

Full check

  • Full pipeline with tests

  • Monitoring dashboard exists

  • Alerting configured

PROD

Complete check

  • Runbooks documented

  • On-call rotation defined

  • Incident response ready

CI/CD Pipeline

MVP+
  • Pipeline runs on every push to main
  • Pipeline runs on every PR
  • Linting runs before build
  • Tests run before deploy
  • Build fails if tests fail
  • Deployments are automated (no manual steps)
  • Pipeline status visible to team

Environment Configuration

MVP+
  • Environment variables separated by environment
  • Production secrets not accessible in dev/staging
  • Environment variable documentation exists
  • .env.example updated with required variables
  • Secrets stored in platform secret manager
  • No secrets in version control

Preview Deployments

MVP+
  • Preview deployments created for PRs
  • Preview URLs accessible to reviewers
  • Preview deployments cleaned up after merge
  • Preview environment has test data (not prod)
  • Preview URLs protected if needed

Rollback Capability

MMP+
  • Previous deployment can be restored quickly
  • Rollback process documented
  • Rollback tested periodically
  • Database migrations are reversible
  • Feature flags allow quick disable of features
  • Rollback doesn't require code changes

Feature Flags

MMP+
  • Feature flag system in place
  • New features deployed behind flags
  • Flags can be toggled without deploy
  • Flag cleanup process defined
  • Kill switches for critical features
  • Percentage rollouts possible

Database Migrations

MMP+
  • Migrations run automatically in pipeline
  • Migrations are backward compatible
  • Rollback migrations exist
  • Migration history tracked
  • No manual database changes
  • Migration tested in staging before prod

Common Issues & Quick Fixes

Manual deployments required: Set up Vercel/Netlify auto-deploy from main branch
Tests skipped to deploy faster: Make test passing a hard requirement for merge
No way to rollback: Use Vercel instant rollback or implement deployment versioning
Secrets in .env committed: Add .env to .gitignore, use platform secrets
PRs deployed to production: Use preview deployments for PRs, main for prod

Incident Response Template

  1. 1. Detect: Alert received, verify the issue exists
  2. 2. Communicate: Notify stakeholders, update status page
  3. 3. Contain: Prevent further damage (feature flag, rollback)
  4. 4. Diagnose: Find root cause using logs and monitoring
  5. 5. Fix: Implement and deploy solution
  6. 6. Verify: Confirm issue resolved, monitoring normal
  7. 7. Review: Post-incident review, document learnings

Tools & Resources

CI/CD Platforms

Error Tracking

Monitoring

Feature Flags

AI Agent Commands

Use these prompts with your AI coding agent to check deployment readiness:

  • Run the Deploy & Observe audit at MVP level
  • Check CI/CD configuration for best practices
  • Find missing error boundaries in the app
  • Audit health check endpoint coverage
  • Check for missing environment variable documentation

Related audits

Data & Security

Security in deployment pipeline

Code & Testing

Tests in CI/CD pipeline

Speed & Performance

Performance monitoring in production