# IT disaster recovery plan template for UK businesses

URL: https://computecloud.co.uk/tools/it-disaster-recovery-plan-template/
Last updated: 2026-11-16

> Free IT disaster recovery plan template for UK businesses on Microsoft 365 and Azure. Word and PDF, no sign-up, with a worked Microsoft 365 outage example.

A simple plan to fill in, for owners, office managers and IT leads at UK businesses on Microsoft 365, with or without Azure or a server on site. Download it as Word or PDF, with no sign-up, or read it below with guidance under every section.

## Key points

- **An IT disaster recovery plan** is a written set of steps for getting your IT systems and data back after a serious incident, such as ransomware, a fire or a long Microsoft 365 outage. It records who declares a disaster, which systems come back first, how long each can be down (RTO), how much data you can afford to lose (RPO), where the backups are, and how to reach staff when email is down.
- **This free IT disaster recovery plan template** covers plan details, roles and contacts, systems and priorities, backups, seven scenarios, a testing log and an after-incident review, with an appendix on Microsoft's SLA. Download it as Word or PDF, with no sign-up.
- **The template includes a filled-in example** of its Microsoft 365 outage scenario, from the Exchange Online outage that began on 12 November 2026 (EX1550543).
- **Test the plan** with a tabletop exercise and a restore test. The NCSC's [Exercise in a Box](https://www.ncsc.gov.uk/section/exercise-in-a-box/overview) has free tabletop exercises, including a ransomware attack.
- **Free to adapt and share:** the template is licensed under CC BY 4.0, so you can change it and pass it on as long as you keep the credit line.

Download the template (free, no sign-up): [Word (.docx), 51 KB](https://computecloud.co.uk/downloads/it-disaster-recovery-plan-template.docx) · [PDF, A4, 73 KB](https://computecloud.co.uk/downloads/it-disaster-recovery-plan-template.pdf). Version 1.0, 16 November 2026.

Free to use and adapt under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). If you share it, credit it as: IT disaster recovery plan template by ComputeCloud (computecloud.co.uk), CC BY 4.0.

## What should an IT disaster recovery plan include?

An IT disaster recovery plan should say who is in charge, what to restore first, where the backups are, and what to do in each incident you're likely to face. This template has eight sections. Use the table as a checklist for your own plan, or to jump to a section.

| Section | What it records |
|---|---|
| [Plan details](https://computecloud.co.uk/tools/it-disaster-recovery-plan-template/#plan-details) | Owner, version, next review date, and where copies are kept, including one outside Microsoft 365 |
| [Roles and contacts](https://computecloud.co.uk/tools/it-disaster-recovery-plan-template/#roles-and-contacts) | Who declares a disaster and who does what; suppliers; where to report an incident; how to reach staff without email or Teams |
| [Systems and priorities](https://computecloud.co.uk/tools/it-disaster-recovery-plan-template/#systems-and-priorities) | Each system, where it runs, what depends on it, its RTO and RPO, and its priority |
| [Backups](https://computecloud.co.uk/tools/it-disaster-recovery-plan-template/#backups) | What's backed up, where, how often and for how long, who can restore it, and the last restore test |
| [Scenarios](https://computecloud.co.uk/tools/it-disaster-recovery-plan-template/#scenarios) | First steps, who acts and when to escalate for seven incidents: a Microsoft 365 outage, ransomware or malware, loss of the office or server room, an internet connection failure, a compromised account or accidental deletion, an Azure virtual machine or region failure, and a key supplier failing |
| [Testing log](https://computecloud.co.uk/tools/it-disaster-recovery-plan-template/#testing-log) | Each test, what happened and what changed afterwards |
| [After an incident](https://computecloud.co.uk/tools/it-disaster-recovery-plan-template/#after-an-incident) | The review, the lessons learned and the updates to the plan |
| [Appendix A: Microsoft's SLA](https://computecloud.co.uk/tools/it-disaster-recovery-plan-template/#microsoft-sla) | Where Microsoft publishes its SLA and how to claim a service credit |

## The template, section by section

Each section starts with how to fill it in. Grey text is an example for you to replace. The Word and PDF files contain the same text.

### 1. Plan details

*How to fill it in:* Say who owns the plan, which version this is and when it will next be reviewed. Keep at least one copy where you can reach it without Microsoft 365: the NCSC advises keeping plans, checklists and contact details available for when you can't get into your computer systems.

| What to record | Example |
|---|---|
| Organisation | [Organisation name] |
| What this plan covers | [Offices, teams and systems covered, for example our office in [town], Microsoft 365, one server in Azure] |
| Plan owner | [Name], [job title] |
| Deputy owner | [Name], [job title] |
| Approved by | [Name], Managing director |
| Version | [1.0] |
| Date of this version | [Date] |
| Next review date | [Date] |
| Who holds a copy | [Plan owner, deputy owner, IT support provider] |
| Master copy kept in | [SharePoint site or folder] |
| Copy kept outside Microsoft 365 | [Printed copy with the plan owner, and a PDF on a phone or a USB stick kept off the network] |

> Keep a printed copy. If Microsoft 365 is down or your network has been attacked, the copy in SharePoint or OneDrive may be the one you can't open.

### 2. Roles and contacts

*How to fill it in:* Name who declares a disaster and who does each job, with a deputy for each. Write down how you'll reach people if email and Teams are down: the NCSC suggests alternative channels such as phone lines or messaging apps, and an offline copy of contact lists for key staff, suppliers and partners.

Roles (example rows):

| Role | Name and deputy | What they do | How to reach them without email or Teams |
|---|---|---|---|
| Declares a disaster and leads the response | [Name], Managing director. Deputy: [Name] | Decides when to invoke the plan, approves spending, signs off messages to customers | [Mobile number] |
| Technical lead | [Name], IT lead, or [IT support provider] | Checks Service health, contains the problem, restores systems and data | [Mobile number] |
| Communications | [Name], Office manager. Deputy: [Name] | Tells staff, customers and suppliers what's happening and when the next update is due | [Mobile number] |
| Recorder | [Name] | Keeps the incident log: what was found and when, decisions taken, by whom and why | [Mobile number] |
| Data protection lead | [Name] | Decides whether a personal data breach must be reported to the ICO | [Mobile number] |

| What to record | Example |
|---|---|
| We invoke this plan when | [For example: a system in priority tier 1 is down for longer than its RTO, any sign of ransomware, or the office can't be used] |
| Who can invoke it | [Managing director, or the deputy if they can't be reached] |

**Reaching staff when email and Teams are down**

| What to record | Example |
|---|---|
| Main channel | [For example: a group text message, or a messaging app group on work phones] |
| Backup channel | [For example: a phone tree, where each manager calls their own team] |
| Who sends the messages | [Name], Office manager |
| Where the staff contact list is kept | [Printed with the plan; exported from the HR system on [date]] |
| First message to staff | [We have a problem with [system]. [What still works]. Next update by [time] on [channel].] |

Supplier contacts (example rows):

| Supplier | What they provide | Account or contract number | Phone | Notes |
|---|---|---|---|---|
| [Internet provider] | Office internet connection | [Account number] | [Fault line] | [Fault reporting hours] |
| [IT support provider] | Support for laptops, Microsoft 365 and servers | [Contract number] | [Phone] | [Support hours and out-of-hours number] |
| [Microsoft licensing partner] | Microsoft 365 and Azure subscriptions | [Tenant name] | [Phone] | [Who raises an SLA claim] |
| [Insurer] | Cyber insurance | [Policy number] | [Incident line] | [What must be reported, and how soon] |
| [Software supplier] | [Accounts or line-of-business software] | [Account number] | [Phone] | [Where our data is held] |

**Where to report an incident** (checked 16 November 2026; check these each time you review the plan):

- **A business under a cyber attack in England, Wales or Northern Ireland:** phone Report Fraud on 0300 123 2040 straight away, as it asks businesses under attack to do. Report Fraud is the new name for Action Fraud and is run by the City of London Police. Other cyber crime and fraud can be reported online at [reportfraud.police.uk](https://www.reportfraud.police.uk/).
- **In Scotland:** Police Scotland isn't part of Report Fraud. Call 101, use its [online form](https://www.scotland.police.uk/secureforms/contact/), or report in person.
- **The NCSC:** report a cyber incident at [report.ncsc.gov.uk](https://report.ncsc.gov.uk/). The NCSC says this doesn't meet any legal or regulatory duty to report, and it won't pass information to regulators without your consent. GOV.UK's [where to report a cyber incident](https://www.gov.uk/guidance/where-to-report-a-cyber-incident) service helps you work out who else to tell.
- **The ICO, for a personal data breach:** if a breach is likely to put people's rights and freedoms at risk, report it within 72 hours of becoming aware of it, where feasible, with whatever you know so far. The clock starts when you discover the breach, not when it happened. [Report a breach to the ICO](https://ico.org.uk/for-organisations/report-a-breach/personal-data-breach/).
- **Your insurer:** as your policy requires. Note its deadline in the supplier table above.

### 3. Systems and priorities

*How to fill it in:* List every system the business needs, starting with the ones it can't work without. **RTO** (recovery time objective) is how long a system can be down; **RPO** (recovery point objective) is how much data, measured in time, you can afford to lose. These are your business's choices, so agree them with the people who use each system: the NCSC says the needs of the business, not only IT, should decide what is restored first.

Systems and priorities (example rows):

| System and owner | Where it runs | What depends on it | RTO (how long it can be down) | RPO (how much data you can lose) | Priority tier |
|---|---|---|---|---|---|
| Sign-in and accounts (Microsoft Entra ID). Owner: [Name] | Microsoft 365 | Everything that uses a Microsoft 365 sign-in | Example: 1 hour | Example: not applicable | 1 |
| Email and calendars (Exchange Online). Owner: [Name] | Microsoft 365 | Quotes, customer contact, Teams calendar and scheduling | Example: 4 hours | Example: 1 hour | 1 |
| Files (SharePoint and OneDrive). Owner: [Name] | Microsoft 365 | Every team's documents | Example: 1 day | Example: 1 day | 1 |
| [Accounts software]. Owner: [Name] | On-premises server | Invoicing and payroll | Example: 2 days | Example: 1 day | 2 |
| [Order system]. Owner: [Name] | Azure virtual machine | Orders and stock | Example: 8 hours | Example: 4 hours | 1 |
| [Phone system]. Owner: [Name] | Other cloud app | Customer calls | Example: 4 hours | Example: not applicable | 1 |
| Office internet connection. Owner: [Name] | Office | Every cloud service, when used from the office | Example: 4 hours | Example: not applicable | 1 |

**Priority tiers** (change them to suit you): **1** restore first; **2** restore once tier 1 is back; **3** can wait until normal work has resumed.

> Write down what each system depends on, such as Microsoft 365 sign-in, the internet connection, a server or a supplier. The NCSC advises identifying which systems, and the systems they rely on, support your critical work and should be recovered first.

### 4. Backups

*How to fill it in:* For each system, record what is backed up, where, how often and for how long, who can restore it and when you last tested a restore. The NCSC advises keeping backups separate from your network, in a different place or a cloud service designed for it, because ransomware targets backups, and checking that you know how to restore.

Backups (example rows):

| System | What's backed up | Where the backup is kept | How often | How long it's kept | Who can restore it | Last restore test |
|---|---|---|---|---|---|---|
| Microsoft 365 | Mailboxes, OneDrive accounts and SharePoint sites | [Microsoft 365 Backup, or a third-party backup service] | [Daily] | [1 year] | [Name], IT lead | [Date]: restored [a deleted folder] in [time] |
| [Accounts software] | [Database and attachments] | [Azure Backup, plus a copy kept offline] | [Nightly] | [30 days] | [IT support provider] | [Date] |
| [Order system] | [The whole Azure virtual machine] | [Azure Backup] | [Daily] | [30 days] | [Name] | [Date] |
| Laptops | [Work files, kept in OneDrive] | [OneDrive] | [As files change] | [See Microsoft's windows below] | [Each user, or IT support] | [Date] |

**What Microsoft keeps if you have no backup.** Microsoft 365 keeps deleted items for set periods so you can undo mistakes. These are Microsoft's windows, not a backup.

Microsoft's retention windows are listed under [what Microsoft 365 recovers](https://computecloud.co.uk/tools/it-disaster-recovery-plan-template/#microsoft-365), below. The Word and PDF files show them in this section.

Microsoft says that for every type of cloud service you own your data and are responsible for protecting it. To go back further than these windows, or to recover from ransomware or a mass deletion, you need a backup: Microsoft 365 Backup, Microsoft's own pay-as-you-go add-on for Exchange, OneDrive and SharePoint, or a third-party tool. Our [Microsoft 365 backup guide](https://computecloud.co.uk/microsoft-365/backup/) compares them.

For servers, Microsoft describes Azure Backup as a way to back up workloads in Azure and on-premises, and says it covers backup and restore, while Azure Site Recovery covers disaster recovery and failover. See [Azure backup and disaster recovery](https://computecloud.co.uk/azure/backup-disaster-recovery/).

> Don't leave backup drives connected to your network, and before you restore, connect backups only to devices you know are clean and scan them for malware. Both are NCSC advice for recovering from ransomware.

### 5. Scenarios

*How to fill it in:* Each scenario has first steps, who acts and when to escalate. Change the steps to fit your systems, and delete any scenario that doesn't apply. Whatever happens, start the incident log straight away: the NCSC advises one record of what was found and when, which decisions were taken, by whom and why.

| What to record | Example |
|---|---|
| Incident log kept in | [A notebook, or a document stored outside Microsoft 365] |

> Use UK time for every entry in the incident log, and note the time zone of anything you copy from a supplier.

#### a. Microsoft 365 outage

*How to fill it in:* Use this when email, Teams or files stop working for several people at once. Check Microsoft first: Microsoft says to check Service health to see whether an issue is already known before you call support or spend time troubleshooting. Our [guide to checking for a Microsoft 365 outage](https://computecloud.co.uk/blog/outlook-down-microsoft-365-outage/) has the detail.

**Before it happens**

- Turn on Service health emails: in the Microsoft 365 admin centre, go to **Health** > **Service health** > **Customize** > **Email**. Each admin can add up to two addresses. Make one an address outside Microsoft 365, so a notice can still reach you if Exchange Online is down.
- Install the Microsoft 365 Admin app on admins' phones. Microsoft says it shows Service health, with push notifications.
- Check offline email is on. Classic Outlook uses Cached Exchange Mode by default, which keeps a copy of the mailbox on the computer; new Outlook for Windows has offline access, on by default (**Settings** > **General** > **Offline**). On Remote Desktop Services, Microsoft says new Exchange accounts use Online Mode by default, so those users have no offline copy.
- In OneDrive, mark the files people would need in an outage as **Always keep on this device**, so they open without a connection.

**First steps**

1. Check **Health** > **Service health** in the Microsoft 365 admin centre (admin.microsoft.com). If you can't sign in, check Microsoft's public status page, status.cloud.microsoft, or @MSFT365Status on X.
2. If an active issue matches what staff see, note its ID and select **Manage notifications for this issue** to get each update by email. If nothing matches, select **Report an issue**: the problem may be in your own setup or network.
3. Tell staff what's affected, what still works and when Microsoft's next update is due, using the channels in Roles and contacts if email or Teams is affected.
4. Keep working offline. Outlook can still show email it has already downloaded. Classic Outlook sends email written offline when it reconnects; new Outlook puts new messages in the Outbox. Files kept on the device still open.
5. Start a timeline for a possible SLA claim: when problems started and ended for you, how many users were affected and where, and the errors they saw. See the appendix on Microsoft's SLA.
6. When Microsoft says the service is restored, check that email is arriving and that queued messages have caught up. Save Microsoft's updates and its post-incident report: Service health keeps 30 days of history.

| What to record | Example |
|---|---|
| Who acts | [Name], IT lead: Service health and Microsoft's updates. [Name], Office manager: messages to staff and customers. |
| Escalate when | [Email is down for longer than its RTO, or Microsoft lists nothing and IT support can't find a local cause within an hour] |

#### b. Ransomware or malware

*How to fill it in:* Use this if files are encrypted, a ransom demand appears, or devices behave strangely. The NCSC's guidance on [recovering from a highly disruptive cyber attack](https://www.ncsc.gov.uk/collection/disruptive-cyber-attacks/recovering) covers the first hours in detail.

**First steps**

1. Decide quickly whether to disconnect affected devices from the network or shut them down. The NCSC says disconnecting can stop attackers sending further commands, limit the spread and preserve evidence, while shutting down can lose evidence. Take specialist advice if you can.
2. Call your IT support provider or a cyber incident response provider. The NCSC lists [assured Cyber Incident Response providers](https://www.ncsc.gov.uk/schemes/cyber-incident-response/find-a-provider).
3. Report it: in England, Wales or Northern Ireland, phone Report Fraud on 0300 123 2040 straight away; in Scotland, call Police Scotland on 101. Report it to the NCSC as well (see Roles and contacts).
4. If personal data may be affected, start the ICO clock: a reportable breach must be reported within 72 hours of becoming aware of it, where feasible.
5. Tell your insurer, as your policy requires.
6. Before you restore, check your backups: when were they last taken, are they still available and have they been compromised? Connect them only to devices you know are clean, and scan them for malware first.
7. Change passwords, starting with admin accounts, once your responder agrees it's safe. The ICO suggests changing passwords as one way to contain a cyber incident.

> The NCSC says paying a ransom doesn't guarantee you get your data back, and law enforcement doesn't encourage, endorse or condone paying. Practise this scenario with the NCSC's free ransomware tabletop exercise in Exercise in a Box.

| What to record | Example |
|---|---|
| Who acts | [Name], Managing director: leads and decides. [IT support provider]: contains and restores. [Name]: reports to the ICO. |
| Escalate when | [Immediately: treat any sign of ransomware as a disaster and invoke the plan] |

#### c. Loss of the office or server room

*How to fill it in:* Use this if a fire, flood or power cut means people can't use the office, or the server room is damaged or without power.

**First steps**

1. Make sure everyone is safe and follow the building's emergency procedures before anything else.
2. Tell staff where to work, from home or [another site], using the channels in Roles and contacts.
3. Check what still works. Cloud services such as Microsoft 365 can usually be reached from home; systems on a server in the building can't.
4. Ask your IT support provider to restore on-premises systems from backup somewhere else, in the order in Systems and priorities.
5. Divert office phones, and update voicemail and the website if customers can't reach you.
6. Tell your insurer, and keep photos and a list of damaged equipment.

| What to record | Example |
|---|---|
| Who acts | [Name], Office manager: staff and building. [IT support provider]: systems. |
| Escalate when | [The office will be unusable for more than [1 day], or the server can't be restored within its RTO] |

#### d. Internet connection failure

*How to fill it in:* Use this when the office loses its internet connection but Microsoft 365 itself is working. Microsoft's SLA doesn't cover a network failure at your site or between your site and Microsoft's data centre, so this one is yours to plan for.

**First steps**

1. Check whether it's just the office: can staff reach Microsoft 365 on a phone using mobile data?
2. Report the fault to your internet provider, and note the reference number and the estimated fix time.
3. Move people to a backup connection, such as [a 4G or 5G router or phone hotspots], or to working from home.
4. Staff can keep working from email Outlook has already downloaded and from files kept on their device.
5. If staff need a server in the office from elsewhere, use [our remote access method].

| What to record | Example |
|---|---|
| Who acts | [Name], Office manager: reports the fault. [IT support provider]: backup connection. |
| Escalate when | [The provider can't fix it within [4 hours], or the phones also depend on the connection] |

#### e. Compromised account or accidental deletion

*How to fill it in:* Use this if an attacker is using someone's account, or email or files have been deleted, by mistake or on purpose. Microsoft's guide to [responding to a compromised email account](https://learn.microsoft.com/en-us/defender-office-365/responding-to-a-compromised-email-account) has the full steps for admins.

**First steps: a compromised account**

1. Disable the account while you investigate, and revoke its active sessions, as Microsoft advises.
2. Remove any sign-in (MFA) methods or devices you don't recognise, apps that shouldn't have access, and admin roles the account shouldn't have.
3. Remove any mail forwarding the attacker set up.
4. When the investigation is finished, reset the password and turn the account back on.
5. If personal data may have been exposed, decide whether to report it to the ICO (see Roles and contacts), and follow the NCSC's advice on [hacked accounts](https://www.ncsc.gov.uk/section/respond-recover/hacked-accounts).

**First steps: deleted email or files**

1. Email: the user can get deleted items back with **Recover Deleted Items** in Outlook or Outlook on the web, within the deleted item retention period (see Backups).
2. Files: restore them from the OneDrive or SharePoint recycle bin. To undo a lot of changes at once, Microsoft lets you restore a whole OneDrive, or a SharePoint document library, to any point in the last 30 days.
3. A deleted user: an admin can restore the user, and their mailbox, from **Deleted users** in the Microsoft 365 admin centre within 30 days.
4. Anything older than Microsoft's windows: restore from your backup.

| What to record | Example |
|---|---|
| Who acts | [Name], IT lead, or [IT support provider] |
| Escalate when | [An admin account is affected, the attacker sent email to customers, or data is beyond Microsoft's windows and has no backup] |

#### f. Azure virtual machine or region failure

*How to fill it in:* Use this if a server you run in Azure stops working, or Microsoft reports a problem with an Azure region. Microsoft says that deploying to a region pair, such as UK South and UK West, doesn't by itself give you disaster recovery or failover: you have to plan it. [Azure backup and disaster recovery](https://computecloud.co.uk/azure/backup-disaster-recovery/) explains the options.

**First steps**

1. Check Azure Service Health in the Azure portal. Microsoft says it gives a personalised view of the Azure services and regions you use, and Resource Health shows the state of an individual virtual machine.
2. If one virtual machine is affected, restore it from Azure Backup, following the steps in Backups.
3. If you replicate with Azure Site Recovery, follow your recovery plan to fail over to the secondary region, then fail back when the primary is running again. Microsoft says recovery plans let you set the order in which machines fail over.
4. Tell staff and customers which services are affected and when you expect them back.

| What to record | Example |
|---|---|
| Who acts | [IT support provider], with [Name], IT lead |
| Escalate when | [The server can't be restored within its RTO, or Microsoft reports a region-wide problem] |

#### g. A key supplier fails

*How to fill it in:* Use this when a supplier you depend on, such as your IT support provider, a software company or a cloud service, stops providing its service. The NCSC advises planning for the sudden loss of a supplier's services and understanding the resilience arrangements of your key suppliers.

**First steps**

1. Find out what has stopped and for how long, using the numbers in Roles and contacts.
2. Switch to your workaround: a manual process, another supplier, or a system you already have.
3. Make sure you can still get your data: where it's held, and how you'd export it.
4. If the supplier has admin access to your systems and may have been attacked, review that access. The NCSC warns that attackers may target suppliers as a way into their customers.
5. Check the contract for notice periods, exit terms and how your data is returned.

| What to record | Example |
|---|---|
| Who acts | [Name], Managing director, with the owner of the affected system |
| Escalate when | [The supplier can't say when service will return, or a tier 1 system is affected] |

### 6. Testing log

*How to fill it in:* Record every test: what you tested, what happened and what you changed afterwards. The NCSC says plans should be tested regularly, with a full exercise where possible, because many plans fail in practice.

Testing log (example rows):

| Date | Test type | Scenario or system | Who took part | What happened | What we changed afterwards |
|---|---|---|---|---|---|
| [Date] | Tabletop exercise | Microsoft 365 outage | [Managing director, office manager, IT lead] | [We couldn't find staff mobile numbers without Outlook] | [Printed contact list added to the plan] |
| [Date] | Restore test | [Accounts server, from Azure Backup] | [IT support provider] | [Restored in [3 hours], within its RTO] | [None] |
| [Date] | Contact test | Reaching staff without email or Teams | [All staff] | [[18] of [20] replied within an hour] | [Two phone numbers updated] |

### 7. After an incident

*How to fill it in:* After every incident, review what happened while it's fresh, and update the plan. The NCSC advises revising your incident plan with the lessons learnt, so the same thing can't happen in the same way again.

| What to record | Example |
|---|---|
| Incident | [What happened, and any supplier or Microsoft incident ID] |
| How long it affected us | [Start date and time] to [end date and time] |
| Review held on, and who attended | [Date], [names or roles] |
| What worked | [...] |
| What didn't work | [...] |
| Reports made | [NCSC, Report Fraud or Police Scotland, ICO, insurer, SLA claim], with reference numbers |
| Changes to the plan | [What changes, who does it, by when] |
| Plan updated to version | [1.1], on [date] |

**Questions to ask**

- Did we know quickly that something was wrong, and who told us?
- Could we reach everyone we needed to?
- Were our RTOs and RPOs realistic?
- Did the backups restore, and how long did it take?
- What would we do differently next time?

> For a Microsoft incident, add Microsoft's post-incident report from Service health to your review. Service health keeps 30 days of history, so save a copy.

### Appendix A. Microsoft's SLA

*How to fill it in:* This explains where Microsoft publishes its service level agreement (SLA) and how to claim a service credit. It's a short summary of Microsoft's document: the SLA itself is what counts.

**Where it's published.** Microsoft publishes the Service Level Agreement for Microsoft Online Services, which covers Office 365, Azure, Dynamics 365 and Intune, on its licensing site: [Service Level Agreements (SLA) for Online Services](https://www.microsoft.com/licensing/docs/view/Service-Level-Agreements-SLA-for-Online-Services). The SLA says the current version is always at aka.ms/CSLA.

**Which version applies.** Microsoft says it won't change your SLA during the initial term of your subscription; when you renew, the version current at renewal applies for the whole renewal term.

| What to record | Example |
|---|---|
| Our subscription renews on | [Date] |
| Who submits a claim | [Name], or [our Microsoft licensing partner] |

**What a credit is.** If Microsoft doesn't meet a service level, you may be eligible for a credit towards part of your monthly fees for that service. Service credits are your sole and exclusive remedy, and Microsoft doesn't award them for lost revenue, operational costs or other losses.

**How to claim.** Submit the claim to Microsoft customer support with all the information Microsoft needs to check it. Microsoft rejects claims without the required information, which includes:

- a detailed description of the incident;
- the time and duration of the downtime;
- the names of the affected resources;
- the number and locations of affected users;
- a description of the errors.

**When.** For Microsoft 365 and other services apart from Azure, Microsoft must receive the claim by the end of the month after the month of the incident: for an incident on 15 February, by 31 March. Azure claims must reach Microsoft within 60 days of the incident. Microsoft says it typically processes claims within 45 days.

**What isn't covered.** The SLA doesn't apply to problems outside Microsoft's reasonable control, such as a network or device failure at your site or between your site and Microsoft's data centre, or to problems caused by not following appropriate security practices. Free services and previews aren't eligible.

> Each service has its own definition of downtime and its own credit levels in the SLA. Microsoft makes a good-faith decision on whether a credit is owed, from the information available to it, and you must be complying with your agreement to be eligible.

## IT disaster recovery plan example: the 12 November 2026 Microsoft 365 outage

What does a filled-in IT disaster recovery plan look like? Here is the Microsoft 365 outage scenario completed for a real incident: the Exchange Online outage that Microsoft began reporting at 17:54 UK time on 12 November 2026, EX1550543. [Microsoft's full timeline of the outage](https://computecloud.co.uk/blog/outlook-down-microsoft-365-outage/) is on our outage post.

This is scenario a, Microsoft 365 outage, filled in as a small business on Microsoft 365 might have recorded the outage that began on 12 November 2026. Microsoft's updates are summarised from its public status page and its Microsoft 365 Status account on X, in UK time (GMT). Entries in square brackets are the business's own, and are examples. Microsoft took the incident off its public status page at 05:42 on 14 November without saying there that it was resolved, so this example has no end time.

| What to record | In this example |
|---|---|
| Incident | EX1550543, with MO1550821 (Microsoft's IDs in Service health) |
| What Microsoft said was affected | Exchange Online, and features of other Microsoft 365 services that depend on it. Users may be unable to reach their mailbox in Outlook on the web, Outlook desktop, Exchange ActiveSync or REST-based apps, and email delivery may be delayed. |
| What still worked, according to Microsoft | Most core Teams features, including calls, meetings, group chats and channel messages. Files stored directly in SharePoint Online could still be opened. |
| Cause, according to Microsoft | An issue introduced during an earlier network capacity upgrade, which surfaced after a recent network change. Microsoft said it was a technical issue, not related to cyber or AI. |
| Plan invoked | [18:10 on 12 November, by the managing director] |

Timeline kept for the incident log and the SLA claim (filled in):

| UK time | What Microsoft said | What we did |
|---|---|---|
| 17:54, 12 Nov | Investigating reports of users having issues accessing Exchange Online mailboxes through multiple connection methods in the Americas (EX1550543) | [Staff report that Outlook won't connect. IT lead checks Service health.] |
| 18:07 | The issue may affect users in any region | [Plan invoked. Group text to all staff: email is down, use Teams chat, next update by 20:00.] |
| 18:35 | Elevated load on part of its infrastructure across multiple regions; details under EX1550543 and MO1550821 | [Turned on Manage notifications for this issue, including an address outside Microsoft 365.] |
| 00:56, 13 Nov | Admins may be unable to load some admin centre pages; core Teams messaging and meetings remain largely working | [Noted for the morning: check status.cloud.microsoft if the admin centre won't load.] |
| 09:10 to 10:01 | Email timeout window extended to 48 hours, to preserve delayed messages | [Customers told by phone that emails may arrive late.] |
| 12:00 | Encouraging recovery progress in Europe | [Spot check: [3] of [12] staff can send and receive.] |
| 17:57 | Underlying issue addressed; a technical issue, not related to cyber or AI | [Staff update: service recovering, delayed email still arriving.] |
| 20:02 | Sign-in and access may recover before email delivery fully catches up | [Asked staff to report any email still missing.] |
| 22:00 | Sustained recovery; monitoring for an extended period, including the backlog of queued email, before considering it resolved | [Last report of a problem from our staff at [time].] |
| 05:42, 14 Nov | Incident taken off Microsoft's public status page, without saying there that it was resolved | [IT lead checks Service health for Microsoft's latest update.] |

| What to record | In this example |
|---|---|
| Who acted | [Name], IT lead: Service health and Microsoft's updates. [Name], Office manager: staff and customer messages. [Name], Managing director: invoked the plan. |
| Workarounds used | [Outlook showed email already downloaded, and replies waited to be sent. Teams chat for internal messages. Files opened from SharePoint, and from OneDrive files kept on devices.] |
| Escalated | [09:00 on 13 November, past our 4-hour RTO for email: phoned our ten largest customers.] |
| SLA claim | Microsoft must receive any claim by 31 December 2026, the end of the month after the incident. [Evidence kept: this timeline, the number and locations of affected users, error messages and screenshots. Claim to be submitted by [Name].] |
| Lessons learned | [Our Service health emails went only to an Exchange Online mailbox we couldn't open. We added an address outside Microsoft 365, and a printed staff phone list.] |

> For incidents with broad impact on many organisations, Microsoft says it delivers a preliminary post-incident review in Service health within 48 hours of the incident being resolved, and a final one within five business days. Add it to your after-incident review.

## What's the difference between disaster recovery and business continuity?

A business continuity plan keeps the whole business running through a disruption: people, premises, suppliers and customers. An IT disaster recovery plan is the part that gets IT systems and data back. ISO 22301 is the international standard that sets the requirements for a business continuity management system; BSI publishes it in the UK as BS EN ISO 22301.

| | IT disaster recovery plan | Business continuity plan |
|---|---|---|
| Covers | IT systems, data, accounts and the connections to them | The whole business: people, premises, suppliers, customers and IT |
| Question it answers | How do we get our systems and data back, and in what order? | How do we keep serving customers while something is broken? |
| Suggested owner | The IT lead or IT support provider, signed off by a director | A director or the business owner |
| Guidance to follow | The NCSC's guidance on recovering from cyber attacks, and Microsoft's Well-Architected Framework for Azure | ISO 22301, the international standard for business continuity management systems |

To turn this template into a fuller business continuity and disaster recovery plan, add the business side. The NCSC suggests that a business continuity plan records which services matter most, how long you can operate without them and which systems support them, with workarounds for when the technology isn't available. It also says to involve people from across the organisation, and any customers and suppliers your critical services depend on.

## How do you test an IT disaster recovery plan?

Test an IT disaster recovery plan in two main ways: a tabletop exercise, where the team talks through a scenario with the plan in front of them, and a restore test, where you restore real data from backup and time it against your RTO. The NCSC says plans should be tested regularly, with a full exercise where possible, because many plans fail in practice.

| Test | What you do | What it shows |
|---|---|---|
| Tabletop exercise | The people named in the plan work through one scenario, step by step. The NCSC's ransomware exercise suggests 60 to 90 minutes with a senior leader, a technical lead, a communications lead and someone taking notes. | Whether roles, decisions and contact details hold up. Microsoft says tabletop drills help a team practise its roles and train new members. |
| Restore test | Restore real files, a mailbox or a server from backup to a safe place, and time it. | Whether the backup holds what you need and comes back within your RTO. The NCSC advises checking that you know how to restore, and testing regularly that backups work. |
| Contact test | Reach every member of staff using only the channels in your plan, without email or Teams. | Whether the backup channel and the contact list are up to date. |
| Failover test | If you use Azure Site Recovery, run a disaster recovery drill. Microsoft says you can run drills without affecting ongoing replication. | Whether servers start in the other region, in the right order. |

**Record the results** in the testing log: the date, the test, who took part, what happened and what you changed. Microsoft's Well-Architected Framework says to treat test results as inputs that improve the plan.

**Use the NCSC's Exercise in a Box.** [Exercise in a Box](https://www.ncsc.gov.uk/section/exercise-in-a-box/overview) is free, and the NCSC says you don't have to be an expert to use it. It has 20 micro and tabletop exercises across 12 topics, each with what you need to plan, run and follow up the exercise. The [ransomware tabletop exercise](https://www.ncsc.gov.uk/section/exercise-in-a-box/ransomware-attack) fits scenario b and downloads as a zip of PDFs, with no account needed (checked 16 November 2026). None of the 13 tabletop exercises is a Microsoft 365 outage, so use the worked example above as your own: share Microsoft's updates one at a time and ask the team what they would do next.

**Get a free review of your disaster recovery plan.** Tell us a little about your plan, whether it's this template or your own, and we'll review it for free, covering Microsoft 365, Azure and on-premises servers. You share the plan itself once we're in touch. We can also help you run a test: a tabletop exercise or a restore test. [Ask for a free plan review](https://computecloud.co.uk/tools/it-disaster-recovery-plan-template/review/)

## What does Microsoft 365 recover, and what does your plan need to cover?

Microsoft recovers its own service. It keeps at least three copies of every Exchange Online mailbox database in separate data centres, which it says protects mailbox data from software, hardware and even data centre failures, and it writes every SharePoint and OneDrive file to two regions. But Microsoft says that for every type of cloud service "you own your data and identities", and that most data loss results from customer actions. For those, it gives you set windows to undo deletions:

How long Microsoft keeps deleted data:

| Kept for | What |
|---|---|
| 14 days | Deleted email, by default. An admin can raise it to 30. |
| 93 days | Deleted files in the OneDrive and SharePoint recycle bin. |
| 30 days | A deleted mailbox, before it is gone for good. |
| 30 days | A leaver's OneDrive, by default (30 to 3,650). |

Your plan needs to cover what Microsoft doesn't:

- **Keeping people working during an outage:** offline Outlook, files kept on devices, and a way to reach staff that doesn't depend on Microsoft 365 (scenario a).
- **Going back further, or undoing ransomware or a mass deletion:** a backup, such as Microsoft 365 Backup, Microsoft's pay-as-you-go add-on, or a third-party tool. Our [Microsoft 365 backup guide](https://computecloud.co.uk/microsoft-365/backup/) compares them.
- **Knowing who can restore, and proving it works:** the backups table and the testing log.

Retention policies and litigation hold keep data for compliance, and are in Business Premium, Office 365 E3 and E5, and Microsoft 365 E3 and E5, but not Business Basic, Business Standard or Office 365 E1: [see which plans include them](https://computecloud.co.uk/tools/microsoft-365-plans/#t=retention-policies,litigation-hold). For servers on site or in Azure, see [Azure backup and disaster recovery](https://computecloud.co.uk/azure/backup-disaster-recovery/), or start with a [free Azure assessment](https://computecloud.co.uk/assessment/).

## Frequently asked questions

### Who should own the IT disaster recovery plan?

One named person, with a deputy: usually a director, or the operations or IT lead. They keep the plan up to date, arrange tests and say who can invoke it. In a small business without IT staff, a manager can own it while the IT support provider fills in the technical sections. The NCSC's guidance on disruptive cyber attacks says the chief executive will normally lead the organisation's response, so make sure the most senior person knows the plan.

### How often should a disaster recovery plan be tested and reviewed?

Microsoft's Well-Architected Framework says a disaster recovery plan should be reviewed regularly, ideally every six months, and updated whenever your systems change or a test finds something unexpected. The NCSC says plans should be tested regularly, with a full exercise where possible, because many plans fail in practice. Neither sets a fixed testing interval, so put your next review date in the plan, and review it after every test and every incident.

### Does a small business need an IT disaster recovery plan?

Yes, if it couldn't work for a day without email, files or its main software. The NCSC says one in two small businesses suffers a cyber incident every year, and advises planning for an attack even if you think it unlikely. A small business's plan can be short: fill in the sections that apply and delete the scenarios that don't.

### What do RTO and RPO mean in a disaster recovery plan?

RTO, the recovery time objective, is how long a system can be down: Microsoft defines it as the maximum acceptable time to restore business operations after a disaster. RPO, the recovery point objective, is how much data you can afford to lose, measured in time. For example, an RPO of 24 hours means a nightly backup is enough, while an RPO of 1 hour needs a backup or copy at least every hour. The figures are your choice: agree them for each system with the people who use it, and record them in the systems table.

### Does Microsoft 365 data need a separate backup?

Microsoft says that for every type of cloud service, "you own your data and identities" and you're responsible for protecting them. It keeps deleted items for set periods, such as 14 days for deleted email by default and 93 days in the SharePoint and OneDrive recycle bins. Its own add-on, Microsoft 365 Backup, is aimed at "a ransomware attack that encrypts large swaths of your data" or "an internal accidental or malicious data deletion or overwrite event". So if you'd need to go back further than those windows, plan for a backup. Our [Microsoft 365 backup guide](https://computecloud.co.uk/microsoft-365/backup/) compares the options.

### Can I adapt and share this IT disaster recovery plan template?

Yes. ComputeCloud's IT disaster recovery plan template is licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/), so you can copy, change and share it for any purpose, including in your business. Keep the credit, "IT disaster recovery plan template by ComputeCloud (computecloud.co.uk), CC BY 4.0", link to the licence, and say if you changed it. Your filled-in plan stays yours: the licence doesn't ask you to publish it.

---

Get a free Azure assessment of what you need to protect and how much loss each system can take: https://computecloud.co.uk/assessment/
