# Focused test plan

## Change and purpose
- Feature / candidate build:
- Business goal and acceptance criteria:
- Release decision this testing supports:
- Test owner / product decision owner:

## Risk inventory
| Failure scenario | Consequence | Priority | Proposed evidence | Owner |
| --- | --- | --- | --- | --- |
| Describe the failure, not just the screen | | | | |

## Scope boundaries
- Included journeys, roles and supported clients:
- Excluded areas and reasons:
- Assumptions requiring confirmation:
- Dependencies or integration partners:

## Execution arrangement
- Environment, configuration and build identity:
- Accounts and permissions:
- Data creation, cleanup and isolation:
- Manual, API and automated checks to execute:
- Accessibility or specialist security review required:
- Evidence storage and sensitive-data handling:

## Completion and reporting
- Entry conditions before execution starts:
- Required checks and evidence before a decision:
- How blocked scenarios and defects are reported:
- Who accepts residual risk and until when:
- Summary owner and delivery date:
- Retest triggers after fixes or configuration changes:

## Planning example — fictional
Change: an organisation administrator can invite a new member.
Highest-impact risks: inviting into the wrong organisation; an ordinary member gaining invitation privileges; accepting an expired invitation.
Proposed checks: a successful invite using isolated test data; cross-organisation and role-boundary API requests; expired and already-used invitation links.
Excluded: billing and unrelated profile editing. Their exclusion must be agreed rather than inferred from a missing test case.
Entry condition: the candidate build is deployed to staging and test email delivery is available.
Completion condition: the invitation scenarios have recorded results, privileged access checks are reviewed, and open defects have a release-owner decision.

Revise the plan when the feature scope changes. Keep a link to the eventual test summary so the intended coverage can be compared with what actually ran.