How to write an IT disaster recovery plan that works
An IT disaster recovery plan is a set of instructions for restoring technology after a serious disruption. It should help someone act when the usual administrator is unavailable, the office cannot be accessed or the company’s email is down. A list of backup products is not a recovery plan.
Start with the work your business must keep doing: taking orders, paying staff, dispatching goods or accessing customer records. Identify the systems, people and suppliers that support each activity. Include cloud services; using Microsoft 365 or a hosted accounting platform does not remove the need to plan for lost access, deleted data or supplier outages.
Keep the scope clear. Disaster recovery restores IT services. Business continuity covers how people keep working while those services are unavailable. Link the two documents so temporary arrangements, such as recording orders manually, support the recovery priorities rather than conflict with them.
Set RTO and RPO using business consequences
Recovery time objective (RTO) is the target time to restore a service after disruption. A four-hour RTO means the business aims to have that service usable within four hours. Define when the clock starts and what ‘usable’ means; a running server is not enough if staff cannot complete transactions.
Recovery point objective (RPO) is the maximum amount of recent data the business can tolerate losing, measured in time. A one-hour RPO means you need a recoverable copy no more than one hour behind the disruption. These are planning targets, not guarantees.
Set targets with the managers who own the work, then ask IT to validate whether the existing systems can meet them. Include detection, decision-making, restoration and business checks in the time budget. Shorter targets usually require more investment and operational discipline.
- For each service, record the maximum tolerable interruption and what happens as downtime extends: missed dispatches, delayed wages or contractual consequences.
- Ask how much information could realistically be recreated. Re-entering a few invoices is different from reconstructing a day’s online orders.
- Document any temporary workaround and its limits. A spreadsheet may help for several hours but become unsafe or unmanageable after that.
Document the recovery sequence, owners and access
Build a service recovery register, not just an equipment inventory. Give every critical service a business owner, a technical recovery owner and a deputy. Record its RTO, RPO, dependencies, recovery method and acceptance checks. Review this register whenever a major application or supplier changes.
Map dependencies before deciding the recovery order. Your sales platform may depend on internet connectivity, identity services, DNS and a database. Restoring the application first will not help if nobody can sign in. Specify which steps can run in parallel and which must wait.
Store a controlled copy of the plan somewhere accessible if your main network and cloud identity service are unavailable. Keep an appropriately secured offline copy, with a review date and named owner. Do not put passwords or recovery codes in a broadly shared document.
- Name who can declare a disaster, authorise recovery spending, contact suppliers and approve the return to normal operations.
- Include after-hours supplier numbers, account references, escalation routes and an alternative communication channel if email is unavailable.
- Document how authorised staff obtain emergency access, including multifactor authentication recovery, without depending on the failed service.
- Write numbered recovery instructions with prerequisites, expected results and clear stop-and-escalate conditions.
Make backups and recovery procedures usable
For each critical system, document what is backed up, where copies are stored, how often they run, retention periods and who checks failures. Include application data, configuration, encryption keys and other components needed for restoration. Check SaaS recovery options explicitly; built-in retention may not meet your requirements.
Protect recovery copies from the same incident that affects production. Consider isolated or immutable copies and separate administrative credentials. Record how to retrieve backups if the normal management platform is unavailable, and verify that the necessary keys and access methods remain available.
Write different recovery paths where the risks differ. A hardware failure may allow a straightforward restore. Suspected ransomware requires containment and investigation before recovery: restoring into a compromised environment can cause another failure. Define who decides that the recovery environment and selected restore point are safe enough to use.
Include the return journey. State how temporary systems will be retired, how records created during the outage will be reconciled, and how you will avoid overwriting newer data. Recovery is not complete until the business has checked its records and resumed controlled operations.
Test the plan without interrupting normal work
Start with a tabletop exercise: gather business and IT owners and walk through a realistic scenario without changing live systems. Try an inaccessible Auckland office, a failed identity service or ransomware affecting shared files. Ask who acts first, how they communicate and what prevents the next step.
Next, restore selected data and applications into an isolated test environment. Prevent test systems from sending customer emails, triggering payments, updating live integrations or synchronising changes back to production. Check capacity and bandwidth beforehand so the test does not degrade normal service.
A tabletop cannot prove restoration speed, and an isolated restore cannot prove every production dependency. Use staged tests to build confidence, reserving live failover exercises for an approved maintenance window with explicit risk controls.
- Agree the scope, business participants, success criteria, stop conditions and rollback steps before testing. Confirm that rollback is actually possible.
- Time the recovery from the agreed starting point. Check the age of restored data against the RPO, not merely whether the backup job succeeded.
- Ask a business user to complete representative tasks and verify records. Technical health checks alone do not prove usability.
- Record failures, missing access and unclear instructions. Assign each issue an owner and due date, then retest the affected steps.
Keep the plan owned, current and ready to use
Choose a review and exercise schedule based on business risk and the pace of change. Also review after major migrations, supplier changes, staff departures and incidents. A test from before your last system migration provides limited assurance about the current environment.
Include a communication checklist: who updates staff, customers, suppliers and leadership; who approves messages; and when the next update is due. For a personal information breach, identify who assesses notification obligations under New Zealand’s Privacy Act 2020 and obtains advice. Do not assume restoring systems ends the incident.
Give leadership a short action list showing gaps between agreed recovery targets and demonstrated capability. Each gap needs an owner, a deadline and either funded remediation or explicit risk acceptance. If a managed IT provider supports recovery, confirm responsibilities, after-hours escalation and the recovery work covered by the agreement. A 24/7 contact point is useful, but it does not replace a tested recovery procedure.
Want this handled for you?
CorTech provides 24/7 monitoring, security operations and help desk support for businesses across Auckland and New Zealand.