Data Governance and Reproducibility

Data analysis in internal audit should be reproducible, proportionate, and defensible. The goal is not only to produce an output, but to preserve enough evidence for another auditor to understand what was requested, received, transformed, tested, and reported.

Confidentiality and proportionality

Before requesting or handling data, document:

Data receipt record

For each dataset, retain:

ItemDescription
Source systemApplication, database, report, or data owner source.
Data ownerPerson or team responsible for the data.
Extract date/timeWhen the data was generated or received.
Population definitionWhat records should be included.
Period coveredStart and end dates.
Filters appliedAny extraction filters or exclusions.
Row countNumber of records received.
Field listColumns received and their definitions.
Transfer methodHow the data was shared.
Storage locationWhere the working copy is stored.

Transformation log

Record every material transformation:

Reproducibility checklist

Before relying on the output, confirm:

Practical minimum standard

For each analysis, keep enough documentation to answer four questions:

  1. What data did we request and receive?
  2. Why was that data appropriate for the audit objective?
  3. What did we do to the data?
  4. Why is the output reliable enough for the intended audit use?