Scenarios 07–12
7 — Unusual file count
Stem: normal file = 5k–10k rows; today = 2 million.
Best answer: Reasonableness signal.
Roles: DQ operations, source owner.
Best action: investigate/RCA and confirm whether spike is legitimate before correction/rejection.
Weaker: reject automatically merely because unusual.
Changed fact: contractual hard max = 10k → explicit validity/requirement breach may also apply.
Source: pp. 431, 433–434.
8 — Call-center speed metric
Stem: agents rewarded only for short calls and routinely skip fields; no quality metric.
Best answer: lack of oversight / embedded-process failure.
Supporting: Governance, culture, process ownership.
Roles: business process owner, Governance, Steward.
Best action: add measurable DQ expectations/controls and align performance management.
Weaker: blame agents and clean data; leaves incentive/process cause intact.
Changed fact: UI blocks submission until required data complete → stronger system-enforced prevention.
Source: pp. 427–428, 437–438.
9 — Schema changed downstream
Stem: source renames/changes meaning of field without downstream notice; warehouse reports impossible values.
Best answer: processing/change-control root cause.
Supporting: Integration, Metadata, change control.
Roles: source owner, integration owner, Steward, DQ analyst.
Best action: trace lineage, update mappings/rules, establish formal change notification/control.
Weaker: clean warehouse rows forever.
Changed fact: mappings are correct but source values wrong → source process quality.
Source: pp. 438–439, 457–458.
10 — Emergency patch
Stem: direct SQL fixes one defect but changes unintended records.
Best answer: fixing-previous-issues cause / uncontrolled correction risk.
Supporting: correction controls, issue management.
Roles: DBA/system owner, Steward, DQ operations.
Best action: restore affected data, analyze cause, use controlled correction/change interface and audit.
Weaker: write another quick patch.
Changed fact: approved audited UI correction → manual correction, but not same risky shortcut.
Source: pp. 440, 458–459.
11 — First look at actual data
Stem: leaders agree data is critical and rules exist, but nobody has examined records.
Best answer: Initial DQ Assessment.
Supporting: profiling, critical data, rules.
Roles: DQ analysts, Stewards, SMEs.
Best action: profile/query, compare with rules, validate findings, establish baseline.
Weaker: remediate based only on anecdotes.
Changed fact: stable recurring baseline already exists → ongoing monitoring may be next.
Source: pp. 442–444.
12 — Null-rate surprise
Stem: profiling shows 15% nulls in MIDDLE_NAME.
Best answer: potential condition, not automatically a defect.
Supporting: Profiling vs Rule.
Roles: DQ analyst, Steward.
Best action: confirm applicable business rule/consumer need before judgment.
Weaker: declare nulls always bad.
Changed fact: field mandatory for a regulated population → nulls in that population are completeness defects.
Source: pp. 443–457.