A product engineer leaves the company after years leading the projects for an entire product line. Their account is deactivated through the standard offboarding procedure, the license is reassigned to a new hire, and a few weeks later a strategic customer requests a review of a specification that existed only in that professional's OneDrive. The team searches SharePoint, the emails exchanged with suppliers, and the shared folders, and no one can say whether that file can still be recovered.
Situations like this happen at companies of every size that have moved email, documents, and files to Microsoft 365 and started treating their data as protected by default. The reasoning seems sound: if the platform belongs to one of the largest cloud providers in the world, someone on the other side surely keeps the copies. That premise conflates two very different things: the cloud being up and the company being able to recover what it has lost.
For the partner, this difference affects the intellectual property built up over years of operation. For the executive, it affects the predictability of costs, deadlines, and contracts. For IT leadership, it raises the question that usually comes up only at the worst possible moment: what exactly do we have as a copy, and how long does it take to bring it back?
What shared responsibility really divides
According to Microsoft's documentation on shared responsibility in the cloud, across all deployment types the customer owns its data and identities and is responsible for protecting both. In the SaaS model (software as a service, ready-to-use applications, the category in which the vendor places Microsoft 365), the published matrix assigns the datacenter, physical network, servers, and operating system to Microsoft, while data, configurations, and user identities remain with the customer.
Microsoft further details what the company always retains: the data, including classification, protection, and compliance with governance requirements; the devices that access the services; and user accounts, from creation to removal. In business terms, the platform commits to keeping the service running, and the company is responsible for what it puts inside it and for who holds the key.
It is in this gap between availability and recovery that the costliest risks to operations are concentrated:
- Accidental or intentional deletion: the wrong folder deleted by mistake, or content removed by someone about to leave the company.
- Synchronized ransomware: ransomware (malicious code that encrypts files and demands a ransom) hits a laptop, and OneDrive sync carries the encrypted files to the cloud.
- Compromised account: whoever breaks into an identity inherits the owner's permissions, including the ability to delete content.
- Employee offboarding: removing the account can take years of accumulated work along with it.
- Retention obligations: contracts, customer audits, and regulatory requirements call for evidence preserved for defined periods.
In manufacturing, these risks have a well-known address. An engineering projects folder in SharePoint, containing technical drawings and customer-approved revisions, is usually synced to the computers of the entire product team. A single infected laptop encrypts the local files, sync replicates the corrupted versions, and engineering can no longer open any drawing. The tooling shop waits for the revision, the line does not receive the updated specification, and the promised batch starts running late, with contractual penalties on the horizon.
In Exchange, if a purchasing manager's account is compromised and the attacker deletes mailboxes to cover up payment fraud, they take with them purchase orders, delivery-date confirmations, and the supplier qualification history that a customer's next audit will require.
Microsoft itself acknowledges the weight of the issue: when presenting its backup service for Microsoft 365, it states that what really matters, beyond the copy, is restoring data to a healthy state quickly, and admits that recovering large volumes from a remote, isolated location can take weeks or even months. The fact that this protection exists as a dedicated service, with policies to configure, shows that it depends on a deliberate decision by the company.
How to turn assumption into proven capability
The first step is to identify what actually sustains operations. NIST (National Institute of Standards and Technology, an international reference for security standards), in its contingency planning guide SP 800-34 Rev. 1, published in May 2010, directs organizations to assess systems and operations to determine contingency requirements and priorities, and provides a BIA (Business Impact Analysis) template. Applied to Microsoft 365, the exercise shows that a production line's active projects folder is worth far more, per hour of downtime, than the internal announcements archive.
With the impact mapped, two numbers turn intention into commitment: RPO (Recovery Point Objective, how much data, measured in time, the business can tolerate losing) and RTO (Recovery Time Objective, the maximum time for the service to come back), defined for Exchange, SharePoint, OneDrive, and Teams. Microsoft's documentation shows that these parameters are choices: its native service offers recovery points every 10 minutes for the previous two weeks for OneDrive sites and accounts, and configurable recovery windows of up to 10 years for OneDrive and SharePoint and up to 2 years for Exchange.
From this, it becomes clear what to demand from any solution or partner. The copy of the tenant (the company's dedicated environment in Microsoft 365) must stay out of reach of a compromised account, the backup needs continuous monitoring that flags failures, and restores must be tested periodically, with the actual time recorded. The offboarding process must preserve the employee's data before releasing the license. A backup and disaster recovery strategy fulfills its role when each of these points can be demonstrated.
5 questions every manager should ask
Where does Microsoft's responsibility end and my company's begin for the data stored in Microsoft 365?
The boundary is drawn in the vendor's own responsibility matrix. Microsoft is responsible for the physical infrastructure, the network and the servers that keep the service available, and the company is responsible for the data, the configurations and the identities, which includes deciding how to protect the content, who accesses it and what happens to an account when the employee leaves.
This division makes Microsoft 365 backup a management decision, with a clearly defined owner. Whoever subscribes to the platform needs to know who is accountable for recovery, which policy is active and when the last restore was tested, and if those answers are slow to surface, the responsibility exists only on paper.
Are the recycle bin, versioning and retention policies enough to recover whatever the company loses?
Each native feature solves one part of the problem. The recycle bin undoes a deletion noticed right afterward, versioning restores an earlier edit of a document and retention policies preserve content for compliance and investigation, valuable mechanisms designed for day-to-day operations and governance.
What's left out is precisely the crisis scenario: the loss discovered too late, the mass encryption of an entire project folder, the permanently deleted account or the attacker with administrative permissions who can change the rules themselves. Quickly restoring an entire site to the exact state of an earlier point in time is a backup capability, and it must be contracted, configured and tested as such.
How much does each day of halted production cost while an encrypted project or a deleted mailbox can't be recovered?
The bill rarely shows up on the IT spreadsheet, because the cost spreads across the business. In manufacturing, it adds up to engineering hours redoing drawings, machines waiting for specifications, overtime to get the schedule back on track, expedited shipping to meet the delivery and, in the worst case, contractual penalties or the loss of a certification. The controllership knows the value of each item, and adding them up per day of downtime is the best argument for defining the RTO.
The reverse question brings the predictability executives look for: how much does it cost, per month, to guarantee that recovery happens within hours? Compared with the cost of a single incident, the investment in protection takes on clear financial contours, with a predictable budget on one side and uncapped exposure on the other.
How much of the technical knowledge I spent years building depends on accounts that no one reviews when someone leaves?
In many companies, the most valuable asset is scattered across the accounts of those who produced it: specifications in an engineer's OneDrive, negotiations in a buyer's email, calculation reports on sites created for a single project. This body of knowledge supports customer service, the defense of quality in an audit and, in an eventual sale or succession, the demonstration of the value built.
When offboarding is limited to removing access and releasing the license, part of this asset can walk out the door with the employee. A mature process preserves the content before any deletion, transfers ownership of what belongs to the company and records where each item ended up, protecting the reputation and the value of what has been built.
Who guarantees that the tenant backup exists, is monitored and is tested, without pulling my team away from the projects that matter?
The in-house IT team knows the business like no one else and, precisely for that reason, is constantly absorbed by user requests, improvement projects and production systems. Checking whether backup jobs completed, investigating failures and running restore tests is disciplined, repetitive work that loses priority whenever something urgent comes up.
An MSP (Managed Service Provider) acts here as backup for in-house IT. In a managed IT servicesmodel, the partner continuously runs and monitors the protection routines, delivers reports on tested restores and escalates alerts to the in-house team, which retains decision-making and gains time for the projects that move the business forward.
Frequently asked questions
Does Microsoft 365 already include backup of the company's data?
Under Microsoft's shared responsibility model, protecting data, accounts and configurations falls to the customer company, while the platform is responsible for the infrastructure and service availability. The recycle bin and versioning help with isolated losses noticed in time. Restoring entire sites, accounts and mailboxes to an earlier point requires a configured and tested backup solution.
What is the difference between a retention policy and backup in Microsoft 365?
A retention policy preserves content to meet compliance, audit and investigation requirements. Backup returns data to a healthy state after a deletion, a ransomware attack or a compromised account. The two functions complement each other, and one does not replace the other.
How often should the company test Microsoft 365 backup restores?
Testing frequency should follow the criticality identified in the business impact analysis, with periodic tests for the most sensitive data and whenever the environment changes significantly. Each test needs to record the actual restore time so it can be compared against the agreed RTO, because a backup that has never been restored offers only an assumption of protection.
Find out in just a few minutes where the protection gaps in your environment are with the Microsoft 365 backup gap assessment.
If you want to confirm what is actually protected in your tenant, Zamak runs a no-obligation Strategic IT Assessment alongside your team.