An audit trail is evidence. It should answer who did what, when, from where and with what result. If it cannot answer those questions under pressure, it is decoration. Many organisations have logging switched on and still cannot reconstruct a sensitive change, because the account was shared, the log was editable, the retention period was too short or nobody reviewed the events. This guide treats those gaps as risks to be managed, not as a technical feature to tick.
Why audit trails fail quietly
Logging usually works during normal operations. The weakness appears during an incident or an audit: the record exists in three formats, the clock on one server is wrong, the administrator who made the change can also delete the evidence, or the relevant events were never captured because the application logged only errors. These failures are quiet because nobody tests the trail until they need it. A useful test is to perform a known sensitive action, then ask an independent reviewer to reconstruct it from the logs alone.
Risk register: seven exposures to discuss
Likelihood and impact are assessed qualitatively here. Replace them with your own evidence and document the reason for each rating.
R1: Shared administrator accounts
Risk: a privileged action cannot be attributed to a person. Likelihood: high where teams share credentials for convenience. Impact: high, because discipline and investigation both fail. Controls: unique named accounts, multi-factor authentication, privileged access management and a ban on shared credentials. Residual: medium until legacy systems that cannot separate accounts are isolated or retired.
R2: Logs that the administrator can edit
Risk: the person with motive and access removes or alters the record. Likelihood: medium. Impact: high. Controls: central collection, append-only or write-once storage, a separate logging account, restricted deletion rights and periodic integrity checks. Residual: medium while a small team holds both system and logging administration.
R3: Collection without review
Risk: events are stored but nobody looks, so detection depends on a customer complaint. Likelihood: high in teams without a security rota. Impact: medium to high. Controls: named review owner, alert rules for privileged changes and failed access patterns, a weekly review and a documented escalation route. Residual: medium until review is measured.
R4: Retention that is too short or too long
Risk: evidence disappears before an investigation closes, or personal data is kept beyond its purpose. Likelihood: medium. Impact: high for legal exposure. Controls: a retention schedule by log type, legal hold, automated deletion and a documented exception process. Residual: low when the schedule is tested.
R5: Secrets and personal data inside log payloads
Risk: passwords, tokens, card data or personal details are copied into a system with broad access. Likelihood: medium to high when request bodies are logged by default. Impact: high. Controls: schema-based redaction, allowlisted fields, data loss prevention, restricted log access and a review of new log sources. Residual: medium until every application is covered.
R6: Time drift and inconsistent formats
Risk: events cannot be correlated because servers disagree on time or use different identifiers. Likelihood: medium. Impact: medium. Controls: time synchronisation, UTC storage, structured formats, correlation identifiers and a documented schema. Residual: low with monitoring.
R7: Vendor access outside the trail
Risk: a supplier changes a system or reads data without a record in your logs. Likelihood: medium. Impact: high. Controls: contractual logging requirements, federated access, just-in-time approval, session recording and a monthly access review. Residual: medium where the supplier controls its own platform.
The control stack
- Identity: named accounts, strong authentication and joiner, mover and leaver processes.
- Privilege: least privilege, approval for elevated access and time-bound sessions.
- Collection: application, database, operating system, network and cloud logs in one pipeline.
- Integrity: append-only storage, separation of duties and integrity verification.
- Detection: alerts for sensitive changes, repeated failures and unusual access locations.
- Review: a rota with evidence that reviews happened and actions were closed.
- Protection: access to the logging platform limited to a small, audited group.
Retention, privacy and legal hold
Keep logs for a defined purpose and period, not forever. Classify what is personal data, minimise payloads and restrict who can search them. Set a legal hold process so that deletion pauses when an investigation or dispute begins. If logs leave Bangladesh or are processed by a foreign vendor, document the transfer and the safeguards. Review the retention schedule with legal counsel, because the applicable requirements depend on your sector and contracts.
Questions for the board
- Which systems produce audit logs, and which sensitive actions are not logged?
- Who can delete or alter logs, and is that ability separated from system administration?
- How quickly could we reconstruct a privileged change from last month?
- What is the retention period, and when was deletion last tested?
- Are personal data and secrets excluded from log payloads?
- How is vendor access recorded and reviewed?
- Who reviews alerts, and what evidence shows the review happened?
Reviewing residual risk
For each exposure, decide whether to mitigate further, accept, transfer or avoid the activity. Record the owner, the review date and the trigger that would reopen the decision. A residual risk accepted because the legacy system will be replaced needs a retirement date and a compensating control in the meantime. Without that, acceptance becomes permanent neglect. Report the register to management quarterly so that accepted risks remain visible.
Audit trails belong to a wider monitoring and hardening programme, not to a single tool. Our managed technology team reviews identity, logging and retention together. The pricing page describes scope and response levels; the third-party security risk guide goes deeper on vendor access.
Sources
- NIST SP 800-92, Guide to Computer Security Log Management, accessed 17 November 2026.
- NIST SP 800-53 Rev. 5, Security and Privacy Controls, accessed 17 November 2026.
- OWASP, Logging Cheat Sheet, accessed 17 November 2026.
- DataGuidance, Bangladesh Data Protection Overview, accessed 17 November 2026.
Responses (0)
No comments yet. Add the first one below.
Leave a response
Your email address will not be published. Required fields are marked *.