InnerWeb
Skip to main content

Cost of missing proof in a data protection audit

A personal data inspection usually starts with the question: who changed what, and when. When a plant has no audit log, the answer has to be rebuilt from emails, copies and people's memories, which in the sample calculation takes several times more hours than reading the audit trail. Below are facts about the InnerWeb audit trail, a calculation with explicit assumptions and a comparison chart.

-93%
hours spent on proof in the sample calculation
Source: sample calculation
95%
share of one table in audit size before its exclusion
Source: card
48
entities added to the audit after reviewing 127 entities
Source: card

Why is proof of a change a cost, not a formality?

In an inspection what counts is not only that the data is correct, but that the plant can show who changed it and on what basis. The GDPR accountability principle means an obligation to demonstrate compliance, not merely to declare it. Without a log, every such proof is a project: people have to be found, dates reconstructed and versions reconciled.

Each hour of such a project is an hour of specialists from the legal, IT and OHS departments who are not doing their regular work. The cost grows with every inspection, employee complaint and internal investigation.

What the InnerWeb audit trail records and what it does not

The audit trail records changes to selected data together with the author and the time: permits, the taxonomy of dangerous work, training results, automation rules and inspection routes. The list of audited objects is public, so the administrator knows which questions the system can answer.

The selection rule is: audited are the data that a person may change. Data recorded only by the system (heartbeat, location history, notification logs, webhook queues) is outside the trail, because its history is itself source data. Sensitive fields, such as the password, the confirmation token and the device push token, are excluded from the trail columns, so the log does not store secrets.

Hours to prove changes within a year

Hours to prove changes within a yearBefore and after chart: 140 hours a year without an audit log and 10 hours with the audit trail.050100150hoursWithout an audit log: Five events a year: 140 hours140 hoursWithout an audit logWith the audit trail: Five events a year: 10 hours10 hoursWith the audit trail
Source: Sample calculation based on the assumptions in the calculation section

Sample calculation: hours with a log and without one

Assumptions: within a year the plant must prove five changes (a permit change, a correction of a training result, a rule change, an inspection route change, a change of the scope of authorizations). Without a log, each one requires collecting evidence from several sources, reconciling versions and writing a description for the inspector. With a log, it requires finding the entry and printing it.

The result depends on the number of events and the team, so treat the calculation as a template to be filled with your own numbers. Regardless of the values, the direction is the same: the cost of proof drops by an order of magnitude when the source is the trail and not a reconstruction.

Pitfalls: time, time zone and log size

The audit trail has recorded time in the Europe/Warsaw zone since October 2026; entries from previous winter seasons are offset by 1 hour. When proving timing, keep this in mind, because the inspector may compare hours with other systems and notice the discrepancy.

The second problem is size. Excluding from the audit the table aggregating time spent in a location reduced the audit, because that table took 6,331 MB, which is 95% of all audit tables, and did not document any human action. A coverage review compared 127 entities with 60 audited ones and covered 48 more; the largest added table had 5,825 rows, so the cost of the audit remained negligible.

The third problem is browsing performance. The audit panel pages records with OFFSET, which for a table of 525 thousand rows means about 318 thousand rows scanned per page. For large tables, paging by key or limiting the depth is recommended.

Audit trail coverage after reviewing entities

Audit trail coverage after reviewing entitiesBar chart: 127 entities, 60 audited before the review, 48 added after the review.050100150entitiesAll entities: 127 entities127All entitiesAudited before the review: 60 entities60Audited before the reviewAdded after the review: 48 entities48Added after the review
Source: Documentation of the audit trail function

How to prepare the plant so the proof is at hand

Proof is cheap only when someone has prepared it before the inspection. Check the list of audited objects against the data categories in the register of processing activities, set the retention period of the log and name the people authorized to read it. Once a quarter, run a test question: pick a change from a month ago and measure how long the answer takes.

Example calculation

Team hours to prove changes within a year (sample calculation)

Assumptions
ParameterValue
Number of changes to prove5 events/year
Hours per event without a log (collecting, reconciling, describing)28 hours
Hours per event with a log (finding, printing, describing)2 hours

Result: Without a log: 5 x 28 = 140 hours. With a log: 5 x 2 = 10 hours. Difference 130 hours, i.e. about 93%.

Note: Sample calculation based on illustrative assumptions; substitute your own number of events and hours. It is not a quote or a promise of the inspection outcome.

Video

Frequently asked questions

Does the audit trail replace the register of processing activities?

No. The register describes the processing, while the trail documents changes to data made by people. Both documents complement each other as evidence of accountability.

Does the log contain passwords and tokens?

No. Sensitive fields, such as the password, the confirmation token and the device push token, are excluded from the trail columns.

Do entries from before October 2026 have a different time zone?

Entries from previous winter seasons are offset by 1 hour from the Europe/Warsaw zone. Take this into account when proving timing.

Is Record and Replay used to supervise employees?

No. Session recording serves regression testing, is switched on with the main switch in the settings, and the list of recordings is visible only to the administrator.

Sources

Let's check how many questions about changes your log can answer

In a demo we will show the audit trail, the list of objects and a test search for a change. Book a conversation with the InnerWeb team.

Let's talk