High-Yield Targets 18–34
18 — Initial assessment vs ongoing monitoring
Assessment establishes a baseline by examining actual critical data against expectations. Monitoring repeatedly checks conformance after controls exist. Profiling may support both.
19 — Prioritization vs improvement goals
Prioritization ranks which opportunities matter most. Goals describe the desired improvement result.
20 — Manage DQ Rules
Capture, approve, version, implement, maintain, and provide feedback on testable expectations. Mature practice embeds rule capture into development/change.
21 — Measure / Monitor
Evaluate rule and aggregate conformance/trends. Threshold breach should trigger a defined response.
22 — Issue management
Standard vocabulary/classification, assignment, escalation, workflow, diagnosis, remediation options, decision, implementation, verification, history.
23 — Data Quality SLA
Operational commitment across important boundaries. Include covered elements, business impact, rules/measures, thresholds, notification, response/remediation deadlines, escalation, and possibly rewards/penalties.
24 — Data Quality Response
Role-appropriate scorecards, trends, SLA metrics, issue status, policy/Governance conformance, and business effects of improvements. Reporting must make condition and action visible.
25 — Profiling vs Root Cause Analysis
Profiling reveals patterns and potential defects. RCA explains why the defect occurs. A null rate is evidence, not cause.
26 — Prevention vs Correction
Prevention stops defects entering/propagating. Correction repairs existing records. Root-cause remediation changes the producing condition so recurrence stops.
27 — Effective metric characteristics
Measurable, business relevant, threshold-based, accountable, controllable/actionable, trendable.
28 — Parse vs Standardize vs Enrich
Parse = decompose/recognize. Standardize = convert to target representation. Enrich = add new context/data.
29 — Correction modes
Fully automated; manually-directed with confidence/human review; manual through controlled interface. Direct production patching is a separate high-risk shortcut.
30 — Implementation pattern
Hybrid generally wins: top-down sponsorship/resources/consistency plus bottom-up discovery and incremental wins.
31 — Readiness
Know pain points, actual data condition, risk, and cultural/technical readiness before scaling.
32 — Cross-Knowledge-Area links
Modeling embeds rules; Metadata stores expectations/results; Master/Reference Data supports domains/identity; Integration is a control point; Governance drives decisions/action.
33 — ISO 8000
Application-neutral/portable data meeting stated requirements; supporting standardization for requesting/verifying quality. Lower priority than core operations.
34 — SPC
Use process measurements/control charts to distinguish common from special causes. Statistical control limits ≠ business acceptance thresholds.
Source anchors: pp. 442–470.