Own instance vs. multi-tenant SaaS: what does it mean for your data?
Own instance and multi-tenant SaaS separate customers and data in different ways. Location, access and exit also depend on set-up, licence and contracts.
Table of contents
What is the difference between an own instance and multi-tenant SaaS?
Own instance and multi-tenant SaaS describe how a solution separates customers and shares resources. The separation can sit in the application, the database or the infrastructure; the labels alone don't tell you where data is processed.
In this guide we compare two simplified examples: a solution with a shared application and data store, and a solution with a dedicated application and database per customer. In practice there are hybrid forms. We can build the solution in your own systems.
In a multi-tenant solution, some resources are shared between customers, while the separation can happen at several layers. In our example of an own instance, the application and database are dedicated to one customer; the physical infrastructure and common services may still be shared. Location, access, rights and exit have to be examined in the specific set-up and contract.
Multi-tenant means that a solution serves several customers while sharing some resources. Own instance is used here for a dedicated application and database per customer. Vendors may use the terms differently, so ask which layers are actually separated.
In the guide we go through practical considerations for the two example models: operations, GDPR, data handling and exit. Use the questions to examine a specific solution rather than deducing its properties from the label alone.
What is the difference, concretely?
As a simplified picture, a shared solution can be compared to a shared office space, and a dedicated application and database to separate offices. The picture doesn't say whether the building, the utilities or other common services are separated; likewise, software can be isolated at one layer and share resources at another.
Multi-tenant: shared resources at different layers
A multi-tenant solution can use a common application and a shared database with customer-specific access, for example via a customer ID. Other solutions have separate databases or applications on shared infrastructure. So ask what is shared and how the separation is enforced.
Logical separation is not in itself insecure, but it requires correct access control and operations. Whether customers share a database, physical storage or other services depends on the set-up. The label multi-tenant is not documentation of a particular data layout.
Own instance: your own running copy
In the guide's example of an own instance, the application and the database are dedicated to one company. Networks, servers or supporting services may still be shared, so an own instance doesn't guarantee physical isolation. The instance can be hosted by the vendor or, if the product and terms allow it, on your own infrastructure.
Dedicated instances can be built on the same standard code and data model. Updates can be rolled out together or gradually depending on the chosen operating model, and several versions may be in use. Standardisation and customer separation are different properties, and both must be described.
Why do vendors choose multi-tenant?
The answer is economics and operations, seen from the vendor's side.
Economies of scale for the vendor
Sharing resources can lower the cost per customer and bring parts of operations and maintenance together. The saving depends on scale, load and isolation requirements. A solution can still consist of several installations and databases, which have to be updated and monitored.
Faster roll-out of new features
Shared software can make it possible to roll out changes to many customers. Roll-out can also be gradual, follow customer agreements or have features enabled for selected groups. Simultaneous updates and access to the same features don't follow automatically from multi-tenant.
Assess the benefits and control in the specific contract
Look into how the operating model affects price, updates and your control over data and integrations. A shared solution can give economies of scale, but the specific customer properties and exit terms must be stated in the product and the contract.
GDPR and data processing agreements: what does an own instance change?
The General Data Protection Regulation applies whichever model you use. When a vendor processes personal data on your behalf, your company is the controller and the vendor is the processor, and the rules then require a data processing agreement. The Danish Data Protection Agency explains the division of roles between controller and processor in its guidance on the division of roles, and the EU's overall requirements for how businesses handle personal data are described in the EU's GDPR overview for businesses.
GDPR doesn't require one architecture or the other. But the architecture affects how easy it is to meet the requirements and how easy it is to document that you have:
An overview of where data lives
As the controller, you must be able to account for where personal data is processed and who has access. In both models you need to map all processing locations, including at model, logging and operations vendors, and examine the relevant contracts. The number of links in the chain and the quality of the documentation depend on the specific solution.
Deletion and access
Data subjects' rights (access, rectification, erasure) must be possible to comply with in practice. A separate database can give a delimited place to check deletion, but backups, logging and sub-processors must be included in the check. In both models you need to examine how deletion is documented across the relevant stores and services.
Access restriction
The fewer systems and people that can reach your data, the easier it is to keep access to a minimum. A dedicated instance can limit exposure between customers; the attack surface also depends on access control, integrations and operations.
Multi-tenant is not in itself incompatible with GDPR. An own instance can make certain delimitations easier to examine, but in both models the documentation has to cover the actual set-up, access and supply chain.
Data ownership and exit: what happens when you want out?
The most overlooked question in any software choice is asked only on the day you want to switch: how do we get out, with everything that is ours?
Exit from multi-tenant SaaS
In a SaaS contract, exit may be limited to particular export formats, and continued access to the software may end with the subscription. That has to be examined specifically: which data, relationships, history and configurations can be moved, and what rights do you get? Multi-tenant architecture alone determines neither the scope of the export nor the right to continued operation.
Exit from an own instance
With an own instance, the starting point is different: the database is separate from other customers', and the contract should ensure that you can have it handed over in its entirety. If the contract and the solution support it, you can get a full database copy for migration. Continued self-operation requires the necessary rights and access to dependent services, including in a self-host set-up.
The licence determines how real the freedom is
The licence and the contract are central to continued operation. An own instance can still depend on the vendor's permission and on external services. Look into the specific rights, restrictions and any time-limited licence changes; access to source code is not in itself a guarantee that the whole solution can be kept running.
Security: isolation vs. shared infrastructure
The security discussion between the two models is more nuanced than either camp usually makes it out to be.
The argument for an own instance
Separate applications and databases can limit the spread of faults between customers. The effect also depends on common services, networks and access control. So look into which parts an attack can reach, rather than assuming that a dedicated instance alone limits the damage.
The argument for multi-tenant
Large SaaS vendors have security teams, round-the-clock monitoring and fast patching processes that few SMEs can match themselves. A well-run multi-tenant platform can therefore in practice be better protected than a poorly maintained own instance. Isolation doesn't help if the instance is running forgotten software with no updates.
The honest conclusion
Security is decided less by the architecture and more by operational discipline. An own instance gives structural advantages: the option of limiting the consequences of faults per customer, and a separation that can be documented for the specific set-up, but only if the instances are kept up to date and run professionally. That argues for instance models where the vendor standardises and maintains all instances centrally, rather than models where each installation gets bogged down in its own version.
Self-host: when does it make sense?
Self-host (running the software on your own infrastructure) is the most far-reaching form of own instance. It gives you control over the part of the solution you run yourself, including operations, access and backups, but data can still be sent to external services.
In return, you also take over the responsibility. Self-host makes the most sense when:
- Data is particularly sensitive or subject to internal/external requirements that it must not leave your own environment.
- You have operations skills: somebody has to be able to update, monitor and back up the system.
- Independence is a strategic goal: you want to be able to carry on whatever happens at the vendor.
For companies without their own operations capacity, a vendor-hosted own instance can be an option. Do look into operational responsibility, access and dependencies, though. Whether self-host is possible depends on the product's installation, licence and support terms; it can't be decided from the label single-tenant or multi-tenant alone.
Standardisation without centralisation
One possible challenge with many dedicated instances is maintenance. Common standard code, documented data models and automated roll-out can support operations. Versions and update times may vary depending on the chosen model.
Such an operating model can be described through three separate considerations:
- The product can be standardised: instances can be built on common code and a common data model. Updates can be rolled out together or gradually.
- The data is separated: each customer has its own database. Access must still be secured, including in any shared infrastructure.
- Operations can be standardised: common procedures and automation can help maintenance. Resource consumption has to be assessed for the specific solution.
In this example model, each company has a dedicated application and database, while the product and operations tooling can be common. How much operations work that saves, and which layers are still shared, has to be examined separately.
Own instance and multi-tenant SaaS compared dimension by dimension
| Dimension | Multi-tenant SaaS | Own instance |
|---|---|---|
| Where does data live? | Can be in a shared or separate database on shared infrastructure | In the example, a dedicated database; physical infrastructure may be shared |
| GDPR documentation | Depends on the vendor's processes and contract chain | Requires documentation for the whole processor chain |
| Exit | Export, migration and continued operation depend on product, contract and licence | Database copy and continued operation depend on contract, licence and technical dependencies |
| Security profile | Depends on isolation, access control and operations at the shared layers | Dedicated layers can limit spread; common services must still be secured |
| Updates | Can be rolled out together or gradually depending on operating model and agreements | Can be standardised and rolled out together or gradually |
| Self-host possible? | Depends on product, licence and support terms | Depends on product, licence and support terms |
| Typical price profile | Low entry price, price per user | Varies; dedicated resources per customer |
Grey areas and hybrid forms: when the label misleads
Reality is rarely as clean as the two models above. The market is full of in-between forms, and labels are used loosely, so it is worth knowing the most common grey areas before you take a vendor's word at face value.
A "dedicated schema" in a shared database
A separate schema or separate tables in a shared database is one form of logical separation. It is something other than a dedicated database, but says nothing on its own about security or exit rights. Ask which layers are shared, how access is controlled, and which export or migration options the contract covers.
Dedicated application, shared data layer
Another in-between form is a dedicated application that uses a shared data layer. Here the separation has to be examined both in the application and in the data layer. GDPR documentation and exit depend on the overall processing, access, technical options and contracts.
Enterprise tiers with a "single tenant" add-on
Some SaaS vendors offer dedicated environments for larger organisations. Look into price and terms in the specific offer. A dedicated environment doesn't in itself give a right to keep running the software without the vendor; that depends on the licence, the contract and technical dependencies.
How to cut through
The grey areas don't mean the labels are useless, only that they must be verified. The three questions that cut through any marketing formulation: Do we have our own database (yes/no)? Can the solution be self-hosted (yes/no)? Do we get a full database copy at exit (yes/no)? The answers clarify different things: data separation, the possibility of running it yourself, and exit rights. An own instance can be vendor-hosted without the option of self-host.
How to assess a vendor: 7 questions about data handling
Whichever model you lean towards, these seven questions quickly reveal how well thought through a vendor's data handling is:
- Where exactly does our data live: in which system, in which country, at which infrastructure provider?
- Do we share a database with other customers? If so: how is the separation implemented, and has it ever been audited?
- Who can access our data at the vendor, and is the access logged?
- What does a full export look like? Ask to see an example: format, scope, relationships.
- What exactly happens on termination? How long does data remain with you afterwards, and how is deletion documented?
- Can the solution be self-hosted: now or later? The answer reveals the architecture's real flexibility.
- What does the licence say about our rights to keep running the system without the vendor?
A vendor that answers all seven clearly and concretely takes data handling seriously. Vague answers to questions 4-7 are the classic warning of a difficult exit.
If you'd like to go deeper into the adjacent material, we have also written about the EU AI Act and what it means for Danish businesses, about data quality and CRM data hygiene and about what CRM automation actually is. All three topics bear directly on the question of where and how the company's data lives. If you're looking more broadly at software strategy, the guide to SaaS sales strategy is a good supplement.
Frequently asked questions
Q: Is an own instance the same as on-premise?
No. On-premise means that the software runs on your own physical servers. An own instance means you have your own dedicated installation with its own database; it can be hosted in the cloud by the vendor or self-hosted, if the product, licence and support terms allow it. On-premise is thus one way of running an own instance, but not the only one.
Q: Is multi-tenant SaaS illegal or in breach of GDPR?
No, the architecture is not in itself in breach of GDPR. In both models you must be able to document the relevant processing locations, access, contracts and deletion. An own instance can give a clearer technical delimitation, but doesn't automatically give simpler documentation or lawful processing.
Q: Won't an own instance quickly become outdated software?
That depends on maintenance. Common standard code and a documented update process can support operations. Roll-out can be done together or gradually, and several versions may be in use. Clarify responsibility, support periods and update terms.
Q: What does an own instance cost compared with multi-tenant?
It varies too much for a general answer: multi-tenant typically has a low entry price and a price per user, while instance-based solutions are priced in other ways, because each customer has dedicated resources. Always compare the total cost over several years (including the price of any exit) rather than the monthly price alone.
Q: Can we move from multi-tenant SaaS to an own instance later?
It may be possible, but it depends on the new product, export options, rights and technical dependencies. Look into data volume, format, relationships and configuration, and test a migration. The name of the architecture or the number of years as a customer doesn't alone determine the size of the task.
Q: How do we know whether a vendor's "own instance" is genuine?
Ask the vendor to describe the separation of application, database, physical infrastructure and common services. Ask separately about self-host and database export, as they depend on the product and the contract. A separate schema in a shared database is logical separation, not a dedicated database; that makes it necessary to examine the specific terms.
Tags: data, security, SaaS, GDPR
Contents
Unsure about the rules for your own set-up?
We look at what you use today and what it requires: documentation, consent and what the EU AI Act means for you. 20 minutes, no commitment.
Book a reviewRelated articles

MFA: multi-factor authentication for your business – how to get started
Learn how to implement MFA securely in your organisation. Guidance on methods, costs and best practices for protecting company data.

English-language invoices: requirements for invoices to foreign customers
Invoice in English to foreign customers legally. Danish regulations, EU and international VAT rules, zero-rating, reverse charge, and automation.

Public Procurement in Denmark: how to bid for government contracts
How Denmark's Public Procurement Act works for suppliers. Threshold values, tender procedures, exclusion grounds and AI solutions for proposal management.