Greenfield does not mean constraint-free. The safest implementation establishes outcomes, ownership and data foundations before building workflows. It launches a thin end-to-end service, proves adoption and then expands through governed releases.
| Phase | Primary output | Gate |
|---|---|---|
| Discover | Outcomes and ownership | Scope agreed |
| Foundation | Data and process model | Design approved |
| Vertical slice | End-to-end service | Acceptance passed |
| Go-live | Supported production | Measures active |
Phase 1: Discover measurable outcomes
- 01Identify executive sponsor, process owners, fulfilment teams and user personas.
- 02Baseline restoration time, reassignment, SLA achievement, channel usage and backlog quality.
- 03Separate mandatory policy or regulatory requirements from team preferences.
- 04Agree success measures and a decision authority for scope conflicts.
| Weak objective | Measurable outcome |
|---|---|
| Improve incidents | Reduce reassignment by 25% |
| Better portal | Move 60% of standard requests to self-service |
| Implement SLAs | Measure response and resolution against agreed calendars |
Phase 2: Build foundations before workflows
- User, group and role model
- Locations, companies and departments
- Service ownership and assignment principles
- Priority matrix and state definitions
- Notification standards
- SLA calendars and pause conditions
- Reporting definitions and data ownership
Phase 3: Deliver a thin vertical slice
Configure one realistic flow from submission to assignment, fulfilment, communication, closure and reporting. A vertical slice exposes missing roles, data and policy assumptions earlier than building every form first.
- 01Select a high-volume, moderate-complexity use case.
- 02Use out-of-box behaviour unless a documented gap blocks the outcome.
- 03Demonstrate with real personas and representative data.
- 04Capture configuration decisions and rejected alternatives.
- 05Expand only after the slice passes acceptance criteria.
Phase 4: Test the operating model
| Test layer | Question answered |
|---|---|
| Configuration | Does each rule produce the expected result? |
| Process | Can teams complete the end-to-end journey? |
| Security | Can each persona see and change only what is allowed? |
| Integration | Do failures, retries and duplicates behave safely? |
| Volume | Does the design work at realistic scale? |
| Adoption | Can users complete the task without project-team help? |
Phase 5: Cutover and hypercare
- 01Freeze approved configuration and reconcile deployment packages.
- 02Complete data migration rehearsal and reconciliation counts.
- 03Publish support model, known issues and escalation routes.
- 04Monitor abandoned submissions, assignment changes, SLA breaches and integration errors daily.
- 05Move improvement items into a governed backlog with owners and outcome measures.
Go-live is the beginning of platform governance. A backlog without ownership becomes delayed customisation; a backlog tied to measurable outcomes becomes continuous improvement.
Explore more implementation-focused ServiceNow and architecture guides.