Skip to main content

COMPEL Glossary / GL-57

Grounding Score

The percentage of generative model outputs whose factual claims can be traced to a verifiable source in the supplied context, computed by a grounding evaluator over a fixed test set.

What this means in practice

A low grounding score indicates hallucination risk; a high score indicates the model is faithfully citing retrieved evidence. Measured continuously on a sampled share of production traffic as well as on release-gate evaluations.

Context in the COMPEL framework

A core Safety metric, paired with hallucination rate. Captured in Evaluate and gated on release in Produce.

Where you see this

Grounding Score is most commonly referenced when teams work across the Produce , Evaluate and Learn stages — especially within the Agent Governance layer . It appears in governance artifacts, assessment instruments, and delivery playbooks wherever COMPEL is operationalized.

Related COMPEL stages

Related domains

Synonyms

hallucination rate , factuality score , faithfulness score

See also

  • Trust & Performance Dimensions — The eight continuous-measurement axes against which every AI transformation is evaluated in COMPEL: Value, Reliability, Safety, Responsibility, Compliance, Security, Sustainability, and Adoption.
  • Jailbreak Resistance — A composite score of an AI system's ability to reject adversarial prompts designed to bypass its safety policies, measured against a fixed, versioned red-team test suite.
  • Responsibility Index — A composite scorecard metric for the Responsibility dimension, combining bias delta, explainability coverage, and human-oversight effectiveness into a single index used in executive reviews.