COMPEL Glossary / quality-gate
Quality Gate
A quality gate is a predefined checkpoint in a development or transformation process where deliverables must meet explicit quality criteria before work can proceed to the next stage.
What this means in practice
In AI, common quality gates include data quality validation before model training, performance benchmarking before deployment, fairness testing before production release, governance review before high-risk system launch, and business validation before scaling. For organizations, quality gates prevent the accumulation of technical and governance debt that occurs when substandard work is allowed to progress unchecked. In COMPEL, quality gates are established during the Model stage as part of the delivery standards framework and enforced during Produce, with the AI-specific definition of done and common quality gate patterns covered in Module 2.4, Article 8 on quality assurance and delivery standards.
Why it matters
Quality gates prevent the accumulation of technical and governance debt that occurs when substandard work progresses unchecked through the development pipeline. Without explicit checkpoints for data quality, performance benchmarking, fairness testing, and governance review, organizations accumulate hidden risks that surface only during production failures or regulatory audits. Well-designed gates catch problems when they are cheapest to fix.
How COMPEL uses it
Quality gates are established during the Model stage as part of the delivery standards framework and enforced during the Produce stage. COMPEL defines AI-specific quality gate patterns in Module 2.4, Article 8, including data quality validation before training, performance benchmarking before deployment, and governance review before high-risk system launch. Each gate must produce documented evidence for the Evaluate stage.
Related Terms
Other glossary terms mentioned in this entry's definition and context.