Testing terminology

Purpose and ownership

This glossary is the source of truth for the eight core testing terms below. The user-journey testing strategy remains authoritative for the project's test layers, catalogue rules, prioritization, evidence rules, and maintenance.

Glossary

Term Definition and confidence boundary Relationship to the strategy taxonomy
Unit testing Checks one isolated rule, parser, policy, or transformation and gives fast confidence in that local behaviour; it does not prove that collaborating components or infrastructure work together. Maps to the Unit layer.
Integration testing Checks cooperating components, services, persistence, transactions, or infrastructure boundaries; it does not by itself prove a complete browser interaction. Maps principally to Integration/service; a request boundary is classified as API/view when that is the behaviour exercised.
End-to-end testing Checks a complete supported path across its significant boundaries and gives confidence that the connected path works; it does not necessarily prove that the scenario is representative of an important user's work. Browser-driven evidence maps to Browser; other end-to-end paths are classified by the behaviour they actually exercise.
User-journey testing Checks a realistic, actor-centred sequence from meaningful preconditions to visible and persisted outcomes; it does not require every assertion to live in one browser test. Describes a scenario scope that may combine Browser, API/view, Integration/service, and Owner acceptance evidence.
Acceptance testing Determines whether agreed stakeholder or owner criteria are satisfied in a production-like context; a green automated end-to-end test alone is not sufficient when infrastructure, external integration, performance, or owner judgement is material. Maps principally to Owner acceptance and may use manual evidence, automated evidence, or both.
Smoke testing Performs a shallow check that essential availability and configuration work after a build or deployment; it does not establish broad behavioural coverage. Safe deployed checks map to Production smoke; smoke intent may also be used for a smaller pre-deployment check.
Regression testing Repeats relevant evidence to show that previously supported behaviour has not broken; confidence is limited to the behaviours and risks represented by the selected evidence. A cross-layer testing purpose, not a separate layer.
Exploratory testing Uses guided human investigation to discover unexpected sequences, risks, or defects; session findings are not repeatable assurance until converted into durable regression evidence where appropriate. Maps to the Exploratory layer.

The strategy's API/view and Stateful/model-based layers remain part of its layer taxonomy. They are not additional primary entries in this bounded glossary.

Key distinctions

  • End-to-end and user-journey testing: End-to-end describes the breadth of the path under test. User-journey describes whether the scenario represents meaningful actor behaviour and outcomes. A test can be both, but neither term guarantees the other.
  • Acceptance and end-to-end testing: End-to-end evidence checks a connected path. Acceptance determines whether agreed criteria are met and may also require owner judgement, production-like infrastructure, external services, or performance evidence.
  • Smoke, regression, and exploratory testing: Smoke checks essential availability shallowly. Regression repeats selected evidence against known behaviour. Exploratory testing searches for risks that the selected evidence may not yet represent, and useful findings should inform later regression coverage.