A backup notification says a job completed. It does not show that your business can recover a usable website. The practical test is a controlled restore into a separate environment, followed by checks of the functions customers and staff rely on. This article focuses on planning that exercise. It is different from choosing a backup frequency or responding to a live incident, and it should not involve experimenting on the production site.
The Australian Cyber Security Centre’s small business cyber security handbook, published and reviewed on 7 October 2026, recommends regular backups and testing restoration. We use that recent guidance as the starting point for an operational exercise. The steps below are practical analysis for a business owner and their authorised IT provider. They do not establish that any particular website is secure or free from malware, and successful restoration is not a security clearance.
Choose a scenario the business can understand, such as recovering from a failed release when the original hosting environment is unavailable. State which website and data are in scope. Identify the person authorised to approve the exercise, the technical operator and the business reviewer. Set a clear condition that the live website, customer notifications and production integrations must remain unaffected. If those conditions cannot be met, change the exercise design before beginning.
Write down the questions you want the exercise to answer. Can the provider locate the required backup? Can it restore files and database content together? Can staff identify which information is missing between the backup time and the simulated failure? These questions produce useful evidence. A vague target such as prove backups work can encourage a demonstration that restores a homepage while overlooking the actual business process.
Record the backup’s creation time, source environment and included components. Ask whether it contains website files, the database, uploaded media and configuration needed for the exercise. Do not assume a hosting snapshot includes external email, a third-party booking platform or every file stored elsewhere. List those dependencies separately. The aim is to understand the boundary of recovery rather than to discover exclusions during an emergency.
Use a copy that the operator is authorised to access and handle. Where customer or staff information is present, agree on protection, access and eventual disposal of the test environment. A restore exercise can create an additional copy of sensitive data. Its convenience does not remove the need to control that copy. Avoid circulating backup archives through casual email attachments or shared folders without an approved handling process.
Have the technical provider establish a separate recovery environment with appropriate access restrictions. Search engines and ordinary visitors should not encounter a second public copy of the business website. The provider should also prevent the restored site from sending real customer emails, processing payments or calling production services during the exercise. A visible test banner helps reviewers, but it is not a substitute for those technical controls.
Review integrations before opening the restored site. A scheduled task, payment callback or marketing connector can behave as though it is still running in production. The exercise plan should explain how each relevant dependency is disabled, substituted or tested safely. Where a premium licence or security mechanism restricts use, respect that restriction and document the resulting test limitation. Do not change credentials or bypass controls simply to complete a demonstration.
The operator should record the restore procedure and any errors without exposing secrets in the exercise report. A site that displays an image of the homepage has not demonstrated application recovery. Check that the restored environment can retrieve content, load assets and navigate between representative pages. Record any missing configuration or unsupported software dependency that required intervention.
Keep timing observations specific. Note when the operator began, when the application became available in the test environment and when business validation finished. These are observed exercise timings, not guaranteed future recovery commitments. Real incidents may involve unavailable suppliers, compromised systems or additional investigation. Treat the exercise as evidence of current capability under stated conditions, and record the conditions alongside the result.
Choose a short set of business tasks rather than clicking every page at random. Open a key service page, navigate to contact information and review how an enquiry would be submitted. Use test modes or approved substitutes for anything that could send an external message or create a transaction. If a real delivery test is outside the authorisation, record that part as untested instead of claiming the entire journey passed.
Ask a business reviewer to check content freshness. A technically successful restore may contain an old price, missing article or outdated service description. Identify the backup age and the changes that would need reconciliation before a real recovery could go live. This is where the exercise connects technical restoration to business continuity: the restored system must be both functional and suitable for customers to use.
A website can be restored correctly while the domain points elsewhere or email continues to fail. Review which DNS records, certificates, email services and external platforms the business would need to recover separately. Document their owners and recovery routes without copying credentials into the report. A named responsibility is more useful than a screenshot of an account page containing sensitive information.
Also identify what the exercise intentionally did not test. You may have used the existing hosting account rather than proving recovery at a different provider. You may have restored a recent copy without testing the oldest retained backup. Those are legitimate boundaries when stated clearly. They become a problem only when the report presents a narrow test as proof of every possible recovery scenario.
Separate failures from improvements. A missing database archive is a direct recovery gap. A suggestion to simplify the runbook may be worthwhile but has a different priority. Assign an owner and a due date to each genuine gap, then retest the affected step after correction. Preserve the original result so the business can see what changed rather than receiving a rewritten report that hides the initial failure.
Finish by confirming how the test environment will be removed or retained securely under the agreed plan. Record the backup identifier, tested tasks, observed timings, unresolved issues and next exercise date. The useful outcome is a recovery process people can repeat with evidence. It is not an assurance that incidents cannot happen, or a reason to make unreviewed changes to the live website.
Hosting support; Backup strategy; Website maintenance.
Source: https://www.cyber.gov.au/publication/small-business-cyber-security-handbook. Source date/context: 7 October 2026. Reviewed 1 November 2026. The practical framework is our analysis, not a quotation or an endorsement by the source organisation.