I85PREP
⌘ K
PMI CD-P study materials
PMI CD-P study resources

PMI CD-P: Testing, Quality and Deployment

PMI CD-P · Work through your exam, one topic at a time.

All guide sections

Testing, Quality and Deployment

Domain 4 - Testing, Quality and Deployment

This domain tests whether you can prove an app works, get it adopted and support it after release. Expect questions on test levels, user acceptance testing, data handling in test, release promotion, rollback and hypercare.

Test at the right levels

  • Component checks for formulas and validations.
  • Integration checks for each connector, especially write operations and permission scopes.
  • UAT with real users running real scenarios against acceptance criteria.
  • Regression after every change; non-functional checks for performance, accessibility and security.
  • Role-based and negative tests: what happens when a record is missing, a user lacks access, or two people edit at once?

Test design and data

  • Trace every test back to an acceptance criterion so coverage is visible.
  • Include boundary values, duplicates, partial approvals and cancelled records.
  • Use realistic but de-identified test data. Never copy production personal data into a test environment.

Release and adoption

  • Promote a packaged solution through dev, test and production; keep configuration separate.
  • Write release notes, train users and communicate what changes.
  • Pilot with one team or region before wide release, with a defined rollback path.
  • Plan hypercare: who answers questions in week one, how defects are triaged, what usage looks like.

Worked example

An invoice app passes 20 UAT scenarios. A duplicate-invoice check fails only when the vendor name is edited after entry, because the check runs on create but not on update. Testers caught it because the scenario specified edit then resubmit. The fix moves validation into a shared rule. The team pilots with two regions for two weeks, measures cycle time against the baseline, and keeps the prior version available for rollback.

Traps that cost marks

  • Happy-path-only testing.
  • Testing as admin only, missing record-level visibility problems.
  • Using production data containing personal information.
  • No rollback plan and no support owner after go-live.
  • Declaring success at deployment instead of measuring adoption and outcomes.