If you offer a Software as a Service (SaaS) application to many organizations, you can configure your application to accept sign-ins from any Azure Active Directory (Azure AD) tenant. This configuration is called making your application multi-tenant. Users in any Azure AD tenant will be able to sign in to your application after consenting to use their account with your application.
If you have an existing application that has its own account system, or supports other kinds of sign-ins from other cloud providers, adding Azure AD sign-in from any tenant is simple. Just register your app, add sign-in code via OAuth2, OpenID Connect, or SAML, and put a
"Sign in with Microsoft" button in your application.
Note
This article assumes you’re already familiar with building a single tenant application for Azure AD. If you’re not, start with one of the quickstarts on the
developer guide homepage.
There are four simple steps to convert your application into an Azure AD multi-tenant app:
- Update your application registration to be multi-tenant
- Update your code to send requests to the /common endpoint
- Update your code to handle multiple issuer values
- Understand user and admin consent and make appropriate code changes
Update registration to be multi-tenant
By default, web app/API registrations in Azure AD are single tenant. You can make your registration multi-tenant by finding the
Supported account types switch on the
Authentication pane of your application registration in the
Azure portal and setting it to
Accounts in any organizational directory.
Before an application can be made multi-tenant, Azure AD requires the App ID URI of the application to be globally unique. The App ID URI is one of the ways an application is identified in protocol messages. For a single tenant application, it is sufficient for the App ID URI to be unique within that tenant. For a multi-tenant application, it must be globally unique so Azure AD can find the application across all tenants. Global uniqueness is enforced by requiring the App ID URI to have a host name that matches a verified domain of the Azure AD tenant.
By default, apps created via the Azure portal have a globally unique App ID URI set on app creation, but you can change this value. For example, if the name of your tenant was contoso.onmicrosoft.com then a valid App ID URI would be https://contoso.onmicrosoft.com/myapp
. If your tenant had a verified domain of contoso.com
, then a valid App ID URI would also be https://contoso.com/myapp
. If the App ID URI doesn’t follow this pattern, setting an application as multi-tenant fails.
Update your code to send requests to /common
In a single tenant application, sign-in requests are sent to the tenant’s sign-in endpoint. For example, for contoso.onmicrosoft.com the endpoint would be: https://login.microsoftonline.com/contoso.onmicrosoft.com
. Requests sent to a tenant’s endpoint can sign in users (or guests) in that tenant to applications in that tenant.
With a multi-tenant application, the application doesn’t know up front what tenant the user is from, so you can’t send requests to a tenant’s endpoint. Instead, requests are sent to an endpoint that multiplexes across all Azure AD tenants: https://login.microsoftonline.com/common
When Microsoft identity platform receives a request on the /common endpoint, it signs the user in and, as a consequence, discovers which tenant the user is from. The /common endpoint works with all of the authentication protocols supported by the Azure AD: OpenID Connect, OAuth 2.0, SAML 2.0, and WS-Federation.
The sign-in response to the application then contains a token representing the user. The issuer value in the token tells an application what tenant the user is from. When a response returns from the /common endpoint, the issuer value in the token corresponds to the user’s tenant.
Important
The /common endpoint is not a tenant and is not an issuer, it’s just a multiplexer. When using /common, the logic in your application to validate tokens needs to be updated to take this into account.
Update your code to handle multiple issuer values
Web applications and web APIs receive and validate tokens from Microsoft identity platform.
Note
While native client applications request and receive tokens from Microsoft identity platform, they do so to send them to APIs, where they are validated. Native applications do not validate tokens and must treat them as opaque.
Let’s look at how an application validates tokens it receives from Microsoft identity platform. A single tenant application normally takes an endpoint value like:
https://login.microsoftonline.com/contoso.onmicrosoft.com
and uses it to construct a metadata URL (in this case, OpenID Connect) like:
https://login.microsoftonline.com/contoso.onmicrosoft.com/.well-known/openid-configuration
to download two critical pieces of information that are used to validate tokens: the tenant’s signing keys and issuer value. Each Azure AD tenant has a unique issuer value of the form:
https://sts.windows.net/31537af4-6d77-4bb9-a681-d2394888ea26/
where the GUID value is the rename-safe version of the tenant ID of the tenant. If you select the preceding metadata link for contoso.onmicrosoft.com
, you can see this issuer value in the document.
When a single tenant application validates a token, it checks the signature of the token against the signing keys from the metadata document. This test allows it to make sure the issuer value in the token matches the one that was found in the metadata document.
Because the /common endpoint doesn’t correspond to a tenant and isn’t an issuer, when you examine the issuer value in the metadata for /common it has a templated URL instead of an actual value:
https://sts.windows.net/{tenantid}/
Therefore, a multi-tenant application can’t validate tokens just by matching the issuer value in the metadata with the issuer
value in the token. A multi-tenant application needs logic to decide which issuer values are valid and which are not based on the tenant ID portion of the issuer value.
For example, if a multi-tenant application only allows sign-in from specific tenants who have signed up for their service, then it must check either the issuer value or the tid
claim value in the token to make sure that tenant is in their list of subscribers. If a multi-tenant application only deals with individuals and doesn’t make any access decisions based on tenants, then it can ignore the issuer value altogether.
In the
multi-tenant samples, issuer validation is disabled to enable any Azure AD tenant to sign in.
Understand user and admin consent
For a user to sign in to an application in Azure AD, the application must be represented in the user’s tenant. This allows the organization to do things like apply unique policies when users from their tenant sign in to the application. For a single tenant application, this registration is simple; it’s the one that happens when you register the application in the
Azure portal.
For a multi-tenant application, the initial registration for the application lives in the Azure AD tenant used by the developer. When a user from a different tenant signs in to the application for the first time, Azure AD asks them to consent to the permissions requested by the application. If they consent, then a representation of the application called a
service principal is created in the user’s tenant, and sign-in can continue. A delegation is also created in the directory that records the user’s consent to the application. For details on the application's Application and ServicePrincipal objects, and how they relate to each other, see
Application objects and service principal objects.
This consent experience is affected by the permissions requested by the application. Microsoft identity platform supports two kinds of permissions, app-only and delegated.
- A delegated permission grants an application the ability to act as a signed in user for a subset of the things the user can do. For example, you can grant an application the delegated permission to read the signed in user’s calendar.
- An app-only permission is granted directly to the identity of the application. For example, you can grant an application the app-only permission to read the list of users in a tenant, regardless of who is signed in to the application.
Some permissions can be consented to by a regular user, while others require a tenant administrator’s consent.
Admin consent
App-only permissions always require a tenant administrator’s consent. If your application requests an app-only permission and a user tries to sign in to the application, an error message is displayed saying the user isn’t able to consent.
Certain delegated permissions also require a tenant administrator’s consent. For example, the ability to write back to Azure AD as the signed in user requires a tenant administrator’s consent. Like app-only permissions, if an ordinary user tries to sign in to an application that requests a delegated permission that requires administrator consent, your application receives an error. Whether a permission requires admin consent is determined by the developer that published the resource, and can be found in the documentation for the resource. The permissions documentation for the
Azure AD Graph API and
Microsoft Graph API indicate which permissions require admin consent.
If your application uses permissions that require admin consent, you need to have a gesture such as a button or link where the admin can initiate the action. The request your application sends for this action is the usual OAuth2/OpenID Connect authorization request that also includes the prompt=admin_consent
query string parameter. Once the admin has consented and the service principal is created in the customer’s tenant, subsequent sign-in requests do not need the prompt=admin_consent
parameter. Since the administrator has decided the requested permissions are acceptable, no other users in the tenant are prompted for consent from that point forward.
A tenant administrator can disable the ability for regular users to consent to applications. If this capability is disabled, admin consent is always required for the application to be used in the tenant. If you want to test your application with end-user consent disabled, you can find the configuration switch in the
Azure portal in the
User settings section under
Enterprise applications.
The prompt=admin_consent
parameter can also be used by applications that request permissions that do not require admin consent. An example of when this would be used is if the application requires an experience where the tenant admin “signs up” one time, and no other users are prompted for consent from that point on.
If an application requires admin consent and an admin signs in without the prompt=admin_consent
parameter being sent, when the admin successfully consents to the application it will apply only for their user account. Regular users will still not be able to sign in or consent to the application. This feature is useful if you want to give the tenant administrator the ability to explore your application before allowing other users access.
Note
Some applications want an experience where regular users are able to consent initially, and later the application can involve the administrator and request permissions that require admin consent. There is no way to do this with a v1.0 application registration in Azure AD today; however, using the Microsoft identity platform (v2.0) endpoint allows applications to request permissions at runtime instead of at registration time, which enables this scenario. For more information, see
Microsoft identity platform endpoint.
Consent and multi-tier applications
Your application may have multiple tiers, each represented by its own registration in Azure AD. For example, a native application that calls a web API, or a web application that calls a web API. In both of these cases, the client (native app or web app) requests permissions to call the resource (web API). For the client to be successfully consented into a customer’s tenant, all resources to which it requests permissions must already exist in the customer’s tenant. If this condition isn’t met, Azure AD returns an error that the resource must be added first.
Multiple tiers in a single tenant
This can be a problem if your logical application consists of two or more application registrations, for example a separate client and resource. How do you get the resource into the customer tenant first? Azure AD covers this case by enabling client and resource to be consented in a single step. The user sees the sum total of the permissions requested by both the client and resource on the consent page. To enable this behavior, the resource’s application registration must include the client’s App ID as a
knownClientApplications
in its
application manifest. For example:
knownClientApplications": ["94da0930-763f-45c7-8d26-04d5938baab2"]
This is demonstrated in a multi-tier native client calling web API sample in the
Related contentsection at the end of this article. The following diagram provides an overview of consent for a multi-tier app registered in a single tenant.
Multiple tiers in multiple tenants
A similar case happens if the different tiers of an application are registered in different tenants. For example, consider the case of building a native client application that calls the Office 365 Exchange Online API. To develop the native application, and later for the native application to run in a customer’s tenant, the Exchange Online service principal must be present. In this case, the developer and customer must purchase Exchange Online for the service principal to be created in their tenants.
If it's an API built by an organization other than Microsoft, the developer of the API needs to provide a way for their customers to consent the application into their customers' tenants. The recommended design is for the third-party developer to build the API such that it can also function as a web client to implement sign-up. To do this:
- Follow the earlier sections to ensure the API implements the multi-tenant application registration/code requirements.
- In addition to exposing the API's scopes/roles, make sure the registration includes the "Sign in and read user profile" permission (provided by default).
- Implement a sign-in/sign-up page in the web client and follow the admin consent guidance.
- Once the user consents to the application, the service principal and consent delegation links are created in their tenant, and the native application can get tokens for the API.
The following diagram provides an overview of consent for a multi-tier app registered in different tenants.
Revoking consent
Users and administrators can revoke consent to your application at any time:
If an administrator consents to an application for all users in a tenant, users cannot revoke access individually. Only the administrator can revoke access, and only for the whole application.
Multi-tenant applications and caching access tokens
Multi-tenant applications can also get access tokens to call APIs that are protected by Azure AD. A common error when using the Active Directory Authentication Library (ADAL) with a multi-tenant application is to initially request a token for a user using /common, receive a response, then request a subsequent token for that same user also using /common. Because the response from Azure AD comes from a tenant, not /common, ADAL caches the token as being from the tenant. The subsequent call to /common to get an access token for the user misses the cache entry, and the user is prompted to sign in again. To avoid missing the cache, make sure subsequent calls for an already signed in user are made to the tenant’s endpoint.