An incident starts two clocks. The technical clock is visible: systems are down, staff cannot work, customers may be waiting. It gets attention because it is urgent and difficult to ignore. The evidential clock starts at the same moment, but often nobody is watching it. By the time the claim form arrives, the log retention window may have closed.
A claim is settled on what you can show, not on what actually happened. A Disaster Recovery plan can help restore services and get you trading again; it cannot recreate evidence that was never captured or has since been overwritten. The most expensive document in an incident is the one nobody wrote down at the time.
That matters well before an insurer asks for a chronology. A regulator may need a prompt assessment, customers may ask what happened to their information, and an auditor may sample the incident file months later. These are separate demands, but they draw on much the same material: records that show what was known, when it was known, what decisions followed and whether the controls in place were working.
Under UK GDPR, a personal data breach that meets the notification threshold must be reported to the Information Commissioner’s Office without undue delay and, where feasible, within 72 hours of the organisation becoming aware of it. The phrase that matters is “becoming aware”. In practice, your ability to establish when that happened depends on visibility, alerts and records. If you cannot tell whether personal data was accessed, you cannot make a confident assessment of the breach or its likely effect. The default may become notifying out of caution, which carries its own cost in time, resource and trust. The ICO’s breach reporting guidance is a useful starting point for understanding the reporting process.
The clock does not wait for a perfect forensic picture. What matters is a prompt, documented assessment, including what is known, what remains uncertain and what is being done to find out. Record the basis for decisions, whether you report or not. A carefully maintained breach record is not an admission of fault; it is evidence that the organisation assessed the situation rather than guessing.
The single most valuable artefact in the incident file is often a timeline written while events are unfolding. It should capture who noticed the issue, at what time, from what evidence, what they did next, who was told, and what was decided and by whom. Reconstructing all of that from memory three weeks later tends to produce a document that reads less like a record and more like an excuse.
The practical arrangement need not be elaborate. A shared document with a timestamp column, a named scribe and an instruction that the scribe does not touch affected systems can be enough to establish a reliable narrative. The NCSC’s incident management guidance is explicit that an incident response plan should be linked to the Disaster Recovery, business continuity and crisis management plans, rather than sitting apart from them. The evidence record belongs in that joined-up response too.

Then comes the awkward arithmetic of logs. Identity sign-in records, firewall and VPN logs, endpoint telemetry, email gateway logs, backup job histories and ERP audit trails may each hold part of the explanation. Two common weaknesses make them less useful than they look: retention windows shorter than the time it takes to detect an incident, and clocks that disagree.
An incident noticed eleven days after it began because the person who would normally spot it was on leave cannot be investigated with logs kept for seven. And a timeline assembled from three systems on three clocks is a timeline an insurer will query. Consistent time settings and a documented approach to comparing timestamps are basic operational hygiene, not forensic theatre.
Backups need similar precision. A successful backup job shows that a process completed; it is not evidence that a recovery worked. Test restores and record what was restored, when, and whether the business could use it. A restore that recovers data but rolls back or loses the records explaining what happened can leave you with working systems and an unanswered question.
A cyber insurance policy is a promise conditional on the controls declared at inception and renewal. When a claim is made, the insurer is not only asking what happened. It is asking for proof that the controls described were actually operating on the day. That proof is often the same material a compliance auditor samples: access records, patching records, backup test records, incident records and training records.
ISO 27001 and Cyber Essentials are useful consultancy domains here, not certificates to display and forget. An ISO 27001 assessor samples evidence of logging, monitoring and incident management as a matter of routine. Cyber Essentials asks questions about supported software, patching and configuration; those answers need records behind them. An organisation that can satisfy an assessor is, almost by definition, an organisation that can satisfy a loss adjuster.
For many UK SMEs, the platform holding the transactional heart of the business is IBM i. Its reliability is well earned; its evidential side is less often examined. The audit journal, QAUDJRN, may contain the records needed to understand security-relevant activity. Whether the entries you would need exist at all can depend on system values set years ago by someone who has since left.
Even where journalling is active, retention deserves close attention. Audit journal receivers may be kept for far less time than a claim or investigation will look back over. Spooled files, job logs and output queues may be purged on schedules nobody has reviewed since the machine was installed. The system continues to run, so the settings remain out of sight. A routine purge can quietly remove the evidence before anyone realises it is needed.
This is the difference between a reliable platform and an evidential platform. The first may keep orders, payroll or finance processing through years of ordinary operation. The second can also help explain who changed what, when and under which authority. Review what is being journalled, how long it is kept, where it is held, whether it is held somewhere other than the machine it describes, and who would know how to read it. “We have logs” is not a complete answer if nobody can locate or interpret them.
An estate with clean journals is one of the strongest continuity assets an SME can hold. One with journals switched off for performance reasons is a machine that can recover your business but cannot explain it. That is a strategic risk worth discussing as part of IBM i Management, not just a storage or housekeeping detail.
A modern cloud core can improve the starting position. Through a Dynamics 365 Jump Start, an organisation can assess whether Dynamics 365 Business Central fits its finance and operational needs. It can provide audit trails and change history that would otherwise need to be designed and maintained, which is a real, underappreciated argument for modernising from an ageing on-premise ERP.
But retention and export are settings and decisions, not magic defaults. A platform that records useful activity for ninety days can answer last quarter’s question and not last year’s. When an incident involves the finance system, the questions are familiar: who changed what, when, and was the change approved? A platform configured to answer them can turn a week of forensic accountancy into an afternoon. A platform whose logs have expired cannot.
Cloud Solutions also bring a distinction that is easy to miss in a recovery exercise: recovering data is not the same as preserving records. Restoring to an earlier point can legitimately roll over the very log entries that explain the incident. Export security and audit logs to a separate location protected by separate credentials, and include that archive in recovery planning. An account whose logs live inside the same tenant as the incident is a witness with an interest in the outcome.
Technology Expense Management is an unexpected but practical part of the evidence pack. Proving what was in place on the day of an incident is a register problem: which devices, licences, circuits, contracts and numbers did the organisation have? Technology Expense Management keeps that register current. It earns its keep twice: first by removing spend nobody can justify, and again by producing the schedule of what existed when the loss adjuster asks. Business interruption calculations also depend on knowing what capacity was actually being paid for.
The execution services around the strategy can strengthen this record. Helpdesk Support gives you ticket timestamps, on a consistent clock, that can form the spine of a credible timeline; a disciplined ticket log is evidence nobody had to write specially. Managed IT Services can provide an arrangement in which someone is contractually responsible for maintaining the evidence pack, rather than merely expected to remember it. Network Solutions and Voice and Connectivity matter in passing but can be crucial: firewall records, remote access logs and call records are frequently among the first things requested, and the first things rotated away.

Before an incident, agree what evidence will be needed and where it will come from. Check retention against a realistic detection window, fix the clocks, name a scribe and a custodian, and test that a restore does not destroy the record. Keep copies of the asset register and relevant policies outside the estate they describe. Then diarise a review, so the pack is opened before the incident rather than afterwards.
For a broader view of the Evestaff group's services, visit evestaff.co.uk.
At Evestaff IT Support and Consultancy, the Fractal IT Director approach starts with strategy set by an IT director, then puts the managed service and bolt-on execution around it. The right evidence is part of that strategy: recovery gets the business back on its feet, and an evidence pack is what stops the incident costing twice.
Join The Discussion