Reference Standards
Quality management in WIA-DATA-007 conforms to a coherent stack of international standards rather than ad-hoc team conventions:
- ISO/IEC 25012:2008 — Data quality model: fifteen characteristics across inherent, system-dependent, and user-context dimensions.
- ISO 8000-61:2016 — Data quality management — Process reference model. Defines plan-do-check-act cycle for data quality.
- ISO 8000-110:2021 — Data quality — Master data: exchange of characteristic data — Syntax, semantic encoding, and conformance to data specification.
- W3C Data Quality Vocabulary (DQV) — RDF vocabulary for expressing quality assessments and dimensions.
Quality Tiers
The reference scoring model maps to the WIA bronze/silver/gold tiers using ISO/IEC 25012 dimensions:
- Gold (90–100%): All five mandatory dimensions (Description, Owner, Column Docs, Tags, Freshness) at full weight; no open data quality issues; lineage complete.
- Silver (70–89%): Four of five mandatory dimensions; open issues acknowledged with planned remediation.
- Bronze (50–69%): Three of five mandatory dimensions; minimum publish bar.
- Below 50%: Asset is hidden from discovery for non-owners until tier rises.
Continuous Quality
Each scheduled scan publishes results as DQV assessments so quality scores travel with the dataset wherever it is harvested. Quality regressions trigger webhook alerts to the owner channel; persistent regressions across two scoring windows demote the tier automatically and notify governance.
Practical Application
Quality work compounds when it is automated. The reference scoring engine runs nightly and emits a DQV assessment per dataset. Three operational rules keep quality from sliding:
- Tier never silently drops. A demotion is announced to the owner channel and gets a five-day grace period before discovery is affected.
- Hot assets get attention first. The scoring engine prioritises the top decile of query volume, so the assets the organisation depends on most are also the assets reviewed most frequently.
- Quality issues are first-class artefacts. Each DQV-tagged issue has an owner, a planned remediation date, and a public status, conformant with ISO 8000-61 plan-do-check-act expectations.
Programmes that follow these rules typically observe gold-tier coverage rising from below ten percent at launch to forty percent or more within twelve months without any quality team headcount increase.