01
Quality depends on the use case
A dataset may support a monthly trend while being too inaccurate for a decision on an individual case. Describe the intended use and acceptable error before assessing quality.
The business owner defines critical fields and reference sources. IT can measure data but cannot define its business meaning alone.
02
Eight practical checks
Do not stop at empty fields. Review the reliability of the entire flow.
- required-field completeness;
- valid formats and ranges;
- cross-system consistency;
- freshness and refresh frequency;
- uniqueness and matching rules;
- provenance and change history;
- representativeness of cases and groups;
- permissions, purpose and retention.
03
Build a trustworthy sample
Include seasons, channels, regions, exceptions, complaints, cancellations and historically difficult cases rather than only clean records.
Protect test data like production data. Use minimisation, anonymisation or synthetic examples when they preserve the objective.
04
Make quality continuous
Set alert thresholds, rejection reports and ownership for correcting the source. Manual output correction without root-cause removal creates hidden cost.
Monitor inputs, decision quality and escalations because a changed form, system or behaviour may degrade performance.
SOURCES
Verified source material
Information was checked on the publication date. Legal obligations should always be assessed for the specific organisation and use case.