Lesson14 min
An incident log that holds up in audit
Why it matters
The incident log serves four successive purposes:
- During the crisis — it gives anyone arriving mid-event a full picture in three minutes.
- In the debrief — it lets you rebuild the real chronology, including what collective memory will have rephrased.
- In audit and regulatory notification — DORA and NIS 2 require documenting the chronology and the impact; the log is the source.
- In litigation — it demonstrates the organisation acted diligently, or its absence suggests otherwise.
What goes in it
| Entry type | Example |
|---|---|
| Fact | 14:12 — Monitoring: total failure of the payment service. Source: tool X, alert no. 4471. |
| Decision | 14:41 — Decision to activate the team. Taken by: J. M., operations director. |
| Action | 14:45 — Asked provider Y for an update within 30 min. Assigned to: R. D. |
| Communication | 15:20 — Holding statement no. 1 published on the status page. |
| Discarded hypothesis | 15:35 — "Morning change" hypothesis discarded: the change was not deployed. |
| Non-decision | 16:00 — Decision not to fail over: estimated resolution time shorter than failover time. |
The last two rows are the ones almost always missing. A discarded hypothesis stops it coming back three times. A decision to do nothing is a decision, and with no trace it will later be read as inaction.
What invalidates a log
- It is written after the crisis, from memory;
- Times are approximate or absent;
- Facts and commentary are mixed;
- Authors are not identified;
- It is corrected after the fact without the changes being visible.
The last point is the most serious. A log is completed afterwards — by adding entries dated at the time of completion — but never rewritten. A corrected chronology loses all evidential value, and the correction itself becomes a problem.
Practice
Keeping an incident log is a skill, and it decays. Two exercises a year are enough to maintain it: by the third, the secretary writes twice as fast and naturally separates facts from hypotheses.
Key takeaways
- Timestamp, source, distinguish fact from decision
- Record decisions to do nothing as well
- A log reconstructed after the fact has no evidential value