Data Quality Contextual Relationships

Data quality is not intrinsic to data, it’s a relationship between data and its intended use. The same dataset can be high-quality for one purpose and low-quality for another.

The Contextual Principle

“Fitness for use” is the core concept. Quality questions must always be:

• Quality for what task?
• Quality for which users?
• Quality under what constraints?

Examples

Same Dataset, Different Quality:

• Customer addresses: High quality for marketing, low quality for emergency response (needs real-time updates)
• Historical sales: High quality for trend analysis, low quality for demand forecasting (distribution shift)

Implications

No Universal Quality Score: Multi-dimensional assessment required Use-Case Documentation: Quality requirements must be specified per application Quality Degradation: As use cases change, quality assessments must be revisited Producer-Consumer Gap: Data producers may not understand consumer quality needs

Governance Response

• Catalog data with intended uses, not just schemas
• Define quality thresholds per use case
• Monitor quality relative to specific applications
• Create feedback loops between consumers and producers

Related: 04-atom—data-quality-dimensions-consensus-gap, 04-atom—data-governance