Data Audit Trails: Compliance and Proof of Work
A data audit trail lets you show later how a number came about. We cover which records to keep and what to watch for compliance. Not legal advice.

This article is not legal or financial advice; for retention periods and obligations, work with accounting, legal counsel and your auditor. An audit trail is the set of records that lets you reconstruct afterwards how a piece of data or a number came about.
Which questions should it answer?
- What: which record, which number?
- From where: which source system and which load did it come from?
- When: when was it extracted, when did it change?
- Who: which user or process did it?
- With which version: which definition, rule and software version was used?
Which records should be kept?
- Load ledger: which file or package, when, how many rows, how many new, how many repeated.
- Raw data: in its form before transformation, with its versions.
- Process events: did the pipeline run, did it fail, with which version?
- Definition changes: when and by whom did the formula of a measure change?
- Access records: who looked at which report or exported which data?
From a compliance view
According to the KVKK's own page, Article 12 of the law puts on the data controller the duty to prevent unlawful access and to carry out, or have carried out, the audits needed to apply the law within its own organization. Records of who accessed what can be part of applying such measures. How long records must be kept depends on the relevant legislation and contracts; this is not verified in this article.
Where Floemdata stands
In the Bronze layer of Floemdata Analytics each record is stored with its source table name, business key, extraction ID, extraction time, load time and the batch file it came from; a load ledger also keeps each file's sha256 digest, row count and how many new and repeated records were added. On the center side, pipeline, error, version and installation events from installations are written to an event table with a timestamp. The access-key record keeps a last-use time. A user-level access log showing which user opened which report, and a definition-change log, are not verified in this article (not verified). Write-back to the source system with a rollback record for every write is at the design stage and on the roadmap.
To get started, write step by step how you would trace one number in your most important report back to its source; the step that is missing is the record you need to keep. Source: KVKK, data security obligations.