A dashboard needs to show who accessed data, not only the data itself
Aliases: personal-data access log · user-facing audit trail · data access history
What it is
A personal data access history shows which person, role, system, or processor read, changed, exported, or disclosed which data class, when, and for what stated purpose. A list of what the system holds is static inventory; access history exposes actual use, separating existence, authorization, and completed behavior.
Why it happens
Access originates in invisible background jobs, support tools, administrators, algorithms, and APIs. A dashboard containing only current values makes ordinary processing and unauthorized inspection look identical. Translating audit events into a readable history creates retrospective accountability and gives anomaly reports a specific time, actor, and object.
Studying it
Seed an account with ordinary automation, support inspection, processor export, and one unexpected access. Ask participants to identify actions, choose what warrants inquiry, and pursue it. Measure anomaly detection, actor attribution, purpose understanding, false alarm, and reporting completion. Reconcile display with an audit source that the same business process cannot silently rewrite; otherwise the test validates a story, not behavior.
Where it stops holding
Naming an individual worker may threaten their privacy or safety; role, team, and a traceable event identifier can preserve accountability. Listing every high-frequency machine read can bury human anomalies, so aggregate with expandable originals. Security detection rules may be withheld without hiding that access occurred.
Applying it
- Log read, modification, export, sharing, and administration with time, actor role, purpose, data class, source, and result.
- Layer human, automated, processor, and anomalous states; aggregate repeated machine events with drill-down.
- Link suspicious records to reason, restriction, and contextual reporting paths.
- Reconcile known audit fixtures in both directions and run tamper drills; fail missing, unattributable, or silently deletable events.