I85PREP
⌘ K
CDDP study materials
CDDP study resources

CDDP: Data Quality and Master Data Management

CDDP · Work through your exam, one topic at a time.

All guide sections

Data Quality and Master Data Management

Data Quality and Master Data Management tests two linked disciplines: measuring whether data is fit for its purpose, and maintaining a single governed version of core entities such as customer, product and supplier. Questions target quality dimensions and their measurement, root-cause correction versus cleansing, matching and survivorship rules, and the MDM implementation styles.

Quality dimensions and measurement

Completeness, validity, uniqueness, timeliness, consistency and accuracy are measured separately - a field can be 100% complete and still wrong. Profiling establishes the baseline distribution, null rates and pattern frequencies before rules are written. Business rules become testable assertions with thresholds and owners; a scorecard reports trend, not just a single pass rate. Fixing records downstream is cleansing; fixing the entry process is correction, the only sustainable option.

MDM styles and the golden record

  • Registry: an index of keys and links; source systems remain authoritative.
  • Consolidation: merged copies, read-only, for reporting.
  • Coexistence: consolidated view plus limited write-back.
  • Centralised: MDM becomes the system of record for the attribute. Matching compares records on standardised attributes using deterministic rules and probabilistic scoring; survivorship decides which source wins per attribute - most recent, most trusted system, or steward override. The output is the golden record, versioned and auditable.

Worked example

Two customer records: "J. Smith, 12 High St, 07700 900123" and "John Smith, 12 High Street, +44 7700 900123". Deterministic matching on name alone fails. Standardise - expand the initial, normalise the address, convert the phone to E.164 - then compare on a composite: exact phone plus fuzzy name above 0.9 yields a match. Survivorship sets legal name from the CRM, phone from billing (higher trust), and address from the most recently updated source. The link is stored with its confidence score and a steward review flag so false positives can be reversed.

Traps that cost marks

  • Treating accuracy and completeness as the same metric.
  • Deleting duplicates without preserving cross-reference keys.
  • Assuming MDM always means centralised ownership.
  • Applying survivorship per record rather than per attribute.