Skip to Content

User directories

Local and enterprise directories, Enterprise Connectors and sync, sign-in, password policy, self-signup, mail domains.

A directory is the container that holds users and decides how they authenticate. Every user belongs to exactly one directory, and the directory governs the available profile fields, the authentication methods, the password policy, the self-service options, and (for synchronized directories) where the users come from.

Directory types

TypeWhere users liveAuthentication
LocalStored locally in the platform.Local password, passkey, and the other configured methods.
EnterpriseSynchronized from Active Directory or LDAP.Delegated to the AD/LDAP source through an Enterprise Connector.

Each directory also has a business type that determines the type of the users it holds:

  • B2E — employees.
  • B2B — customer (external business) users, used with organizations.
  • B2C — consumer (external individual) users.

The B2B and B2C directories belong to the Customer Identity add-on; see People → User types.

The Directories list

The list is split into two sub-views: Local (directories stored in the platform) and Enterprise (synchronized AD/LDAP directories). Each row shows a user count that links straight to those users in People. A limit banner appears on the Enterprise sub-view as you approach the number of enterprise integrations your subscription includes.

Note: The built-in default directories are protected: their name and reserved mail domains are fixed and they cannot be deleted. (A directory that still holds users cannot be deleted either — move or remove its users first.)

Creating a directory

  1. Open Directories and choose to create one.
  2. Enter a name and the business type.
  3. To make it an enterprise directory, enable enterprise integration and pick the integration (Active Directory / LDAP).
  4. Save. Local directories open on their settings; enterprise directories open on the Enterprise tab, which lists the connectors bound to the directory and links to Enterprise Connectors so you can enroll one.

Connectors & synchronization (enterprise)

An enterprise directory reaches your Active Directory or LDAP through one or more Enterprise Connectors — lightweight connectors you install in your network that hold the actual connection. The platform never connects to your directory service directly.

Connectors

Connectors are managed centrally on the Enterprise Connectors page, not inside the directory. To connect a directory to its source, create an authentication connector there and bind it to this directory; you can bind several for resilience. The directory's Enterprise tab shows the bound connectors and their online state and links to that page. See Enterprise Connectors for how to create and enroll one.

Synchronization

On the Synchronization tab you control how users flow between the source and the platform:

  • Enable sync — pull users from the directory on a schedule (create and update accounts).
  • Upstream sync — additionally write selected profile changes made here back to the source.
  • Password writeback — write passwords set here back into the source. This is what makes the password of a synced user changeable through the platform at all: the self-service reset, the change in the account portal, the administrative password change and a forced change at sign-in. It is independent of Upstream sync, so profile changes and passwords can be written back separately. With it switched off, passwords are managed exclusively in AD/LDAP.
  • Missing-user action — what to do when a previously synced user disappears from the source: suspend, delete if unused, or delete.
  • Sync deactivated users — whether deactivated users from the source are brought across.

The tab also shows a read-only directory structure overview — the organizational units and containers discovered for users and groups — so you can see exactly which part of the source tree the sync covers.

How records are matched

When a record arrives from a sync, the platform matches it to an existing account by its stable directory identifier (the source object's own id), falling back to the login identifier / email:

  • A record whose identifier matches an existing synced account updates that account.
  • A record with no match is created as a new user.
  • If a new record's email or login identifier already belongs to a different account, the create is rejected as a conflict rather than silently creating a duplicate. The conflict surfaces as an issue so you can resolve it (most often a mail-domain or identifier clash).

Synchronization constrains some lifecycle actions (for example, when a synced user may be deleted). The user detail page explains why an action is unavailable in those cases. See also User lifecycle.

Authentication methods

The directory's Authentication tab decides how its users prove who they are.

Password

Available on local directories and on enterprise directories that allow profile editing. The password policy is configured here:

SettingNotes
Minimum lengthBetween 8 and 64 characters. 8 is the platform minimum, 12 the default for a new directory. NIST recommends 15, which the Security Health report notes as long as the directory stays below it. It is always enforced. A password may be at most 72 characters long; accented letters and other non-ASCII characters take up more than one of them.
Additional composition rulesFour independent switches: require lowercase, uppercase, numbers, special characters. All off by default, in line with current guidance.
HistoryNumber of previous passwords that cannot be reused.
ExpiryForce a change after a number of days (off by default).
BlocklistReject common and breached passwords (on by default; add your own entries, one per line, e.g. your company or product names), plus two pattern switches: block repeated characters (aaa) and block sequential characters (1234, abcd).
Disallow personal dataReject passwords containing the user's own profile data.

The password policy is per directory, so your workforce directory and a customer directory can hold different rules.

Next to Configure sits Test password. It opens a scratchpad field that lists the rules as they are stored and ticks them off as you type, so you can see what the policy asks of your users before they do. Nothing entered there is saved.

How the composition rules work

Each rule is a separate switch, not a step on a scale. Requiring numbers does not also require uppercase, so a policy like "letters and numbers, no special character" (PCI DSS 8.3.6) is one you can actually configure.

Length is the control that carries the policy: a longer minimum protects better than any rule about character types. Raise the length first, and the composition rules only where something obliges you to. That is what they are for. PCI DSS 8.3.6 requires digits and letters, BSI IT-Grundschutz expects character classes as well. Where no such obligation exists, current guidance (NIST SP 800-63B) advises against them: they mostly produce predictable substitutions (Passwort becomes P@ssw0rt!) rather than passwords that are harder to guess. That is why they are all off by default.

Two things are independent of the composition rules. The blocklist (dictionary, breach corpus, repeated and sequential characters, personal data) applies whenever it is switched on. And changing the policy never invalidates existing passwords: the new rules take effect the next time a user sets one.

For an enterprise directory the Enterprise Connector inside your network evaluates the policy as well. Where the connector receives passwords end-to-end encrypted it is the only place that can, because the password never exists in readable form on the platform. Without that encryption the platform validates the password too, and both sets of rules apply.

An connector that predates the independent rules understands only the older cumulative scale, so it applies the nearest set that covers what you configured, which can mean it demands a character type you did not ask for. The console tells you which rules that affects, and the password hint shown to your users lists the rules that will actually apply to them.

Turning passwords off (password-optional directories)

The Password switch itself can be turned off, so the directory's users never see a password screen and sign in with an emailed code instead. Three conditions must be met before the switch can be moved:

  • The directory is a local directory (passwords of an enterprise directory are owned by AD/LDAP and cannot be switched off here).
  • Its business type is not B2E (passwords cannot be switched off for an employee directory).
  • Email code login is already enabled for the directory (see below).

The switch is additionally refused while the directory holds active users without a verified email address: without a password, the emailed code would be their only way to sign in, and verifying an address requires being signed in, so they could not recover on their own. The confirmation dialog shows how many users are affected. It also warns when users hold only an authenticator app or SMS as a second factor, because after the switch only a passkey or security key can raise a session to a higher authentication level; those users keep signing in, but applications that require multi-factor authentication become unreachable for them until they enroll a passkey.

Switching passwords off adjusts the dependent settings automatically:

AreaEffect
Self-signupRequire email verification is enforced and can no longer be turned off.
Password resetThe reset surfaces are switched off, including the self-service reset.
Password expirySet to none.
Signup formNo longer asks for a password.
Admin user creationThe Activate now and set password option is gone; only the activation-link path remains.
Activation linkConfirms the account instead of asking for a password.
Admin user view and end-user portalThe password rows and actions are hidden.
API user creationAn active user is rejected without a verified email address, because the email code would be their only way in. Create the user as provisioned and let the activation link confirm the address, or supply an address already marked as verified. The same applies when activating a user or moving one into the directory.

The switch is reversible. Existing password hashes are kept, but they cannot be used for sign-in while the directory has passwords off. Switching the method back on restores them without a reset.

Note: The password-policy checks in Security Health report not applicable for a directory that has passwords off, so a passwordless directory does not count against your score.

Passkey

If platform passkeys are enabled for the instance, you can allow passkeys (passwordless platform authenticators such as Windows Hello, Touch ID, or Face ID) for the directory, and choose the challenge mode — autofill, an explicit button, or both — and whether passkey enrollment is suggested progressively.

Kerberos (Windows SSO)

Note: Kerberos-based Windows single sign-on requires the kerberosSSO feature, available on the Advanced tier and above.

For enterprise directories you can enable Kerberos so domain-joined Windows clients sign in seamlessly. You enable the method and upload the keytab and configure the service name.

The keytab holds the long-term keys of the Active Directory service account, so it is never displayed again after upload: the configuration dialog shows only its size. Saving the dialog without choosing a new file keeps the stored keytab; the delete button next to it removes the keytab and switches Windows SSO off.

Email code login

Note: Email code login requires the CIAM (Customer Identity) add-on on the Essential plan or above. Without it the switch shows an Add-on badge.

Email code login lets a user sign in with a one-time code sent to their verified email address, either in addition to the password or, with passwords switched off, as the only method. It is aimed at consumer and customer directories where a password is more friction than protection.

Employee directories do not offer it, and the card is not shown there. Three reasons: an employee account keeps a password in any case, so the method would add convenience and no capability; the mailbox usually sits behind this very platform (Microsoft 365, OWA), which makes it circular exactly when it is needed, such as on the first day or after a lockout; and for a directory connected to Active Directory or LDAP it would bypass the directory as the deciding authority, so an account disabled there could keep signing in until the next synchronization catches up. For employees, use a passkey where you want to reduce password prompts.

The directory's email attribute must be enabled on the Attributes tab; while it is not, the switch stays disabled.

To enable it, open the directory's Authentication tab and turn Email code login on.

What the user sees

With the method on and a password still in place, the password screen gains a second button, Email me a sign-in code. It is offered only to users who can actually reach an acceptable assurance level that way:

  • Their email address is verified.
  • They have no second factor enrolled, or their enrolled factor is a passkey / security key.

Users whose only second factor is an authenticator app or an SMS code keep signing in with their password, because neither factor can raise the assurance level of an email-only sign-in.

The button is also absent where the password is exactly what the code cannot stand in for: when the platform asks the user to confirm a sensitive self-service change (their password, email address, or username), and when an application's policy has already sent them to the password screen because an emailed code does not reach the level it requires.

The code screen shows the identifier the user entered, with a Change link so a typo can be corrected without restarting from the application.

The email contains both a six-digit code and a sign-in link, and either one completes the sign-in:

PropertyValue
Validity5 minutes
Codes in flightOnly the most recently sent code works
Attempts per code5; the code is then void and the next screen sends a fresh one
Send rateLimited per user and per IP address

Following the link in the browser the sign-in was started in completes the sign-in, without a further click. A mail scanner or link preview cannot do the same: it has no sign-in session, and the link itself verifies nothing, so opening it has no effect at all.

By default the link works only in that browser. Opened anywhere else it shows a page explaining that and asks the user to type the code into the original window instead. Configure next to the method lifts that restriction, see below.

Each sent code is recorded in the audit log as Email Sign-In Code Sent.

Confirmation across devices

Configure next to Email code login opens Allow confirmation on another device, off by default. With it on, the sign-in link may be opened on a different device than the one the sign-in was started on, typically the phone that received the mail:

  1. The waiting screen shows a three-digit code next to the code field.
  2. Opening the link on the other device shows what is asking to be let in: application, IP address, location, browser and time of the waiting device.
  3. That code has to be entered there. Three wrong entries end the cross-device path for this sign-in; the code typed into the original window keeps working.
  4. On a match, the waiting screen continues on its own within a few seconds.

The page also offers I did not start this sign-in. That invalidates the sign-in code immediately, so nobody is signed in, and records Cross-Device Sign-In Denied in the audit log. A successful confirmation is recorded as Cross-Device Sign-In Approved, together with the IP address, location and browser of the waiting device.

Consider before enabling it. The three-digit code means the confirming person has to be looking at the screen that started the sign-in, which is what stops a link from being forwarded and confirmed by someone else. It does not stop a caller who keeps the user on the phone and reads the code to them. Where sign-in scams targeting your users are a realistic concern, leave the option off: the sign-in link then works only in the browser that started the sign-in, and a code read out over the phone is the only remaining path. Turning Email code login off turns this option off with it.

Assurance

An email-only sign-in is the lowest level in the assurance model, below the single-factor level: a mailbox is not an accepted authenticator under NIST SP 800-63B, so no NIST assurance level is claimed for it. Practical consequences:

  • An application whose authentication policy requires more still asks for a second factor, and only a passkey can satisfy that requirement without a password.
  • The Admin Console always requires more, so it can never be reached with an email code alone.
  • Email is never accepted as the second factor of a sign-in it already started: a code cannot both begin and complete the same login.

See Authentication levels for the full model, and Multi-factor authentication for the separate email code used as a second factor.

Profile schema and attributes

The directory's Attributes / Profile schema tab controls, per field, whether each standard attribute is enabled, serves as a login identifier, is required, and appears at signup. At least one identifier must always be active. This is also where you assign custom attributes to the directory; see User profiles & attributes.

Some standard fields carry a sub-toggle:

  • Full name can include a middle name.
  • Mobile can offer a second mobile number.

Note: On the default employee directory, email cannot be removed as the login identifier — it stays required.

Self-service and self-signup

The directory's settings control what users can do for themselves:

  • Modify profile — let users edit their own profile.
  • Reset password — let users run the password-reset flow. Not offered when the directory has passwords switched off; the emailed sign-in code already covers the "I can't get in" case. For an enterprise directory it additionally requires Password writeback, because otherwise there is nowhere to store the new password.
  • Recovery email — let users register a recovery address. A recovery address only ever serves as the delivery channel for a password reset, so it is switched off and greyed out alongside Reset password when the directory has passwords switched off, or when an enterprise directory has no password writeback. Both switches show why they cannot be changed there.
  • Delete profile — let users delete their own account (local directories).
  • Security checks — show the account-portal security nudges (on by default for external directories, off for employees).
  • Self-signup — allow new users to register themselves (local directories only). Custom attributes marked for signup are collected during registration. With passwords switched off, the signup form no longer asks for a password.
  • Require email verification — when self-signup is on, require new self-service registrations to confirm their email address before the account is created and can sign in (on by default). See Self-service signup for the full flow. A directory with passwords switched off always verifies the address, since it is the only way in. The setting is enforced there and cannot be turned off.

Mail-domain reservation

A directory can reserve one or more email domains. A reserved domain belongs exclusively to that directory: no other directory may claim the same domain, and every path that creates a user or changes an email address honours the reservation. This prevents, for example, a self-signup or social login from seizing a domain that belongs to a managed corporate directory. If a domain is already reserved elsewhere, the operation is rejected with a clear error.

Connect a directory to an organization

A B2B directory can be linked to an organization in its settings (Connect to organization), so accounts in that directory belong to the named business customer.

Account lockout

A directory's password sign-in is protected by the instance-wide account-lockout control: after a configured number of consecutive failed attempts on one account, that account is locked. Depending on the configuration it either unlocks automatically after the window passes or stays locked until an admin (or the user, via password reset) clears it. You configure the threshold under Threat protection; the resulting locked state and the admin unlock action live on the user detail page.

Note: Session idle and absolute timeouts are not a directory setting — they are instance-wide under Settings → Login. See User lifecycle → Sessions.

Lifecycle rules

The directory carries the settings that feed account lifecycle behaviour — such as the missing-user action on sync and the deprovision-to-delete grace period — described in User lifecycle.

Troubleshooting

  • A directory sync fails or users are missing. Sync and provisioning failures are surfaced as issues. Open the issue for the specific error, then check that the enterprise connector is connected and that the attribute mapping covers the fields the sync expects.
  • Sync created duplicate accounts. Records are matched on a stable key. If that attribute changes upstream, or two source records share it, the match no longer lines up and a separate account is created. Confirm the matching attribute is stable and unique in the source.
  • Sign-in against an enterprise directory fails. Password sign-in and sync run through the directory's authentication connectors and automatically use the best-connected one. If every connector for the directory is offline, sign-in and sync for that directory cannot proceed until an connector reconnects. Affected users see a message that the directory is currently unavailable (not a wrong-credentials hint), the attempt is recorded as FAILED_LOGIN_DIRECTORY_UNAVAILABLE and does not count towards their lockout. Check connector status under Enterprise Connectors.
  • A new user is refused with a mail-domain error. The email's domain is reserved by a different directory. Each domain can be claimed by only one directory. Move the reservation or use an address in an unreserved domain.
  • The sign-in link says it only works in the browser the sign-in was started in. That is by design: the link is bound to the browser session that requested it, so a code opened in a mail client on another device cannot complete a sign-in nobody there started. The user goes back to the window where they entered their identifier and types the six-digit code from the same email.
  • A user is still asked for a second factor although email code login is on. The application's authentication policy requires more than the email-code level, so a second factor is demanded on top. Only a passkey satisfies that without a password. Lower the requirement in the authentication policy or the application's own settings if an email code should be enough for that application. The Admin Console is deliberately exempt and always requires more.
  • The Password switch cannot be turned off. It is only movable on a local directory whose business type is not B2E and which already has email code login enabled. The switch's tooltip names the missing condition.
  • Email code login cannot be turned on. Either the subscription does not include it (the switch shows an Add-on badge) or the directory's email attribute is not enabled on the Attributes tab.

See also

On this page