
A backup report that says “successful” is reassuring, but it is not proof that your business can recover. When a server fails, a Microsoft 365 account is compromised, or an office loses connectivity, your team needs working systems, clear decisions and the right access – not a document that has never been used.
This disaster recovery testing checklist is designed for UK small and mid-sized businesses that need practical assurance without turning a test into a major IT project. The aim is simple: establish whether you can restore the systems that matter, in the order the business needs them, within a timeframe everyone understands.
Start with the business impact, not the backup software
Recovery testing often starts with a technical question: can we restore a file, virtual machine or database? That matters, but it is only part of the answer. First decide what an outage means for your operation.
Talk to the people responsible for finance, customer service, operations and compliance. Identify the systems they cannot work without, what they would do manually for a short period, and what would cause unacceptable financial, legal or customer impact. For some businesses, email can be unavailable for a few hours while the line-of-business system must be restored immediately. For another, access to cloud files and telephony may be the priority.
Set two practical targets for each critical service. The recovery time objective, or RTO, is how quickly the service needs to be available. The recovery point objective, or RPO, is how much data you can afford to lose. An RPO of four hours means a restore from last night’s backup may not be acceptable, even if the restore itself works.
These targets should be agreed by the business, not guessed by IT. Faster recovery and tighter data-loss tolerance usually require more investment in backup frequency, storage, standby infrastructure and support cover. That is a sensible trade-off when it protects a genuinely critical service, not when every system is automatically labelled urgent.
Disaster recovery testing checklist: prepare the test
Before testing, make sure the exercise has an owner, a clear scope and authorisation from the right people. A recovery test should produce evidence and improvements, rather than create avoidable disruption.
Use this checklist before the test begins:
- Confirm the critical systems, their recovery order, agreed RTOs and RPOs.
- Choose a realistic scenario, such as ransomware affecting a file server, accidental deletion of Microsoft 365 data, failure of an internet connection, or loss of a cloud-hosted application.
- Name the people who will make decisions, carry out technical recovery, communicate with staff and record timings.
- Check that backup locations, encryption keys, administrator accounts, licences and supplier contacts are available without relying on the failed system.
- Decide whether the test will be isolated, such as restoring data to a separate environment, or whether it will involve controlled live failover.
- Tell affected colleagues what is being tested, when it is happening and what they should do if they spot an issue.
A first test does not need to simulate a complete business shutdown. In fact, starting with a limited restore is often the better approach. It lets you validate the fundamentals safely, then build towards a scenario involving several systems, staff communications and alternative ways of working.
Test recovery, not just restoration
Restoring data is not the same as recovering a usable service. A database may be technically restored but inaccessible to staff because permissions have not returned, the application server cannot connect, or multi-factor authentication depends on an unavailable identity service.
For each system in scope, record the point at which the test started and work through the full recovery path. This should include locating the correct backup set, starting the restore, rebuilding or accessing the required environment, reconnecting dependencies and allowing an authorised user to complete normal work.
Ask a member of the relevant department to validate the result. A finance user should be able to open the accounts system, find expected records and process a representative task. A customer service colleague should be able to locate information and communicate with customers. Technical checks alone can miss the details that stop a team from operating.
Where possible, test different recovery levels. A single file restore proves one thing. Restoring a whole server, a Microsoft 365 mailbox, a SharePoint site or a line-of-business database proves something else. Your programme should reflect the failures most likely to affect your organisation and the consequences if they occur.
Check the dependencies people forget
Most recovery delays come from dependencies rather than the primary system. A file server may be available, but remote staff cannot reach it because the VPN is down. A cloud application may be working, but nobody can sign in because the identity platform or administrator account is unavailable.
During the test, check access from the places your team actually works: office, home and, where relevant, a second site. Confirm that DNS, internet connectivity, firewalls, VPNs, Wi-Fi, telephony and remote access work as expected. If your organisation has multiple offices, establish whether staff can continue from another location if one site is unavailable.
Also verify identity and security controls. Can nominated administrators access recovery platforms using secure, protected accounts? Are break-glass accounts documented, reviewed and stored securely? Can staff complete multi-factor authentication if their usual device is lost? These controls must remain in place during recovery. Removing them to make a test easier can create a different and more serious risk.
Third-party suppliers need attention too. Record their support routes, escalation arrangements and any recovery commitments they have made. A test may show that your internal response is prompt but a key application provider requires information you do not hold, or only supports recovery during limited hours.
Test communications and decision-making
A technical recovery can be successful while the wider response fails. Staff may not know whether to keep working, customers may receive inconsistent messages, and senior leaders may not know who is authorised to approve a switch to contingency arrangements.
Include communications in the exercise. Test how you would contact staff if email or Teams were unavailable. Confirm that contact lists are current and accessible offline. Prepare straightforward messages for employees, customers and suppliers, but do not assume every incident needs the same wording. A short planned outage, suspected cyber incident and confirmed data breach require different decisions and, potentially, specialist legal or regulatory advice.
The incident lead should practise making decisions based on incomplete information. They need to know when to escalate, when to invoke disaster recovery, who can approve additional costs, and when to tell the business that a service is safe to use again. Clear responsibilities reduce the pressure on a single individual and stop recovery work being delayed by uncertainty.
Record evidence, gaps and timings
Do not judge a test by whether it felt smooth. Record what happened. Note the start and finish times for each task, the recovery point achieved, who took part, evidence of successful user validation and any workarounds used.
Then compare the actual outcome with the agreed RTO and RPO. If a restore took six hours against a four-hour target, that is useful information, not a failure to hide. The next question is why. It could be insufficient bandwidth, slow storage, an undocumented dependency, missing access rights or an unrealistic target.
Create an action plan with an owner and due date for each issue. Typical actions include changing backup schedules, protecting additional Microsoft 365 data, documenting application dependencies, adding alternative connectivity, improving administrator access arrangements or updating supplier contacts. Re-test the affected area once the change is complete. A risk is not closed because it has been written down.
Set a testing rhythm that matches your risk
There is no single correct testing frequency. A business with frequent changes to cloud services, acquisitions, new sites or a high exposure to downtime should test more often than one with a stable, low-impact environment. As a baseline, test critical restores at least annually and after major changes. Run lighter checks more regularly, such as confirming backup completion, reviewing alerts and testing individual file or mailbox recovery.
Cyber recovery deserves particular attention. Ransomware can affect live systems and connected backups, so test whether recovery copies are appropriately protected and whether you can restore into a clean environment. Your process should also include checks for signs of compromise before restored systems are returned to use.
A managed IT partner can coordinate the exercise, carry out technical restores and provide an independent record of results. However, business leaders still need to own priorities, acceptable downtime and the decision to invest where gaps are found. Recovery is a business capability supported by technology, not a task handed entirely to IT.
The most useful test is the one that leaves your organisation better prepared than it was the day before. Choose one credible scenario, involve the people who would be needed in a real incident, and turn the findings into completed actions. That is how a recovery plan becomes something your business can rely on when normal operations are interrupted.



