Sign-in methods in your tenant
Each tenant chooses how its users sign in, from passwords to passkeys, and how strictly each method is rolled out.
Authentication settings
Open Authentication in the tenant sidebar. Five methods are available: passwords, email sign-in links, email one-time codes, authenticator apps, and passkeys or security keys. Each method has a rollout: Off, Optional, or Required. The page also shows how many users have set up each method.
Method settings include how long email links and codes last, whether a link may be opened on another device, the name shown in authenticator apps, and whether passkeys need a PIN or biometric, which kind of authenticator to prefer, and whether a passkey alone can sign someone in. The preferred kind is a preference, not a guarantee: the browser offers that kind when a user adds a passkey and reports which kind it used, but the report is not signed and some browsers leave it out. You can also decide when to ask for a second step, whether users who hold management roles always need one (off by default), how many days a browser can be trusted to skip it, how many recovery codes each user gets, and whether people may approve their second step from another signed-in browser (see push approvals). Registration settings control whether people can create their own accounts on your sign-in page and whether users can change their own email address.
Changing these settings needs the Authentication update permission, which the built-in Tenant Admin role has and custom roles can include. Every change is recorded in the tenant's Audit. Settings that would leave users without a way to sign in are refused.
Optional and required methods
An Optional method appears on each user's Security page, where users set it up whenever they choose; it is never offered during sign-in or registration. Only a Required method or a required second step asks someone to set something up while signing in. A Required method is a forced migration: new users set it up while registering, and existing users sign in the usual way one last time, then set it up before they can continue.
A Required method can have a deadline. After the deadline, users who have not set up the method cannot sign in. An administrator with the reset permission can open the user's row in Users and choose Reset sign-in methods. That removes the user's methods, signs them out of every browser, ends the tokens applications hold for them and gives them seven days to sign in and set up again. The user receives an email about the reset.
A Required sign-in method can also replace passwords. Set passwords to Optional or Off first. Each user's password is then removed as soon as they set up the new method.
Second steps
An authenticator app, passkeys and email codes can be second steps, and recovery codes stand in for a lost app or passkey. A second step is always a different method from the first, so the passkey used to sign in can never be its own second step. A password is something the user knows, and an authenticator app, recovery code or passkey is something they have. An email link or code shows only that someone can read the mailbox, which is not an authentication factor, so an email code can be a second step but never makes a sign-in multi-factor, and it cannot follow an email sign-in. An authenticator app can follow a passkey used without a PIN or biometric, so a stolen key alone is not enough, but because both are things the user has, that sign-in is not recorded as multi-factor. A passkey that checks a PIN or biometric proves two kinds at once and needs no second step.
When a second step is required but none of the user's methods can follow the way they signed in, they set up one that can before continuing. If no such method is turned on, a user who has a password is asked to sign in with it instead. The same applies when someone's only second step is the passkey they just used without a PIN or biometric: a user with a password is asked to sign in with it, and the key then serves as the second step, so the key alone is never enough. Someone without a password sets up a method that can follow, when one is turned on. Requiring a second step for users who hold management roles has no effect until an authenticator app, passkeys or email codes are turned on, and the Authentication page says so beside that setting. Turn on an authenticator app or passkeys when the second step should make sign-ins multi-factor.
Wrong second-step codes count for the account across every sign-in attempt. After too many, the second step pauses until 24 hours have passed since the first wrong code counted, or until an administrator resets the user's sign-in methods. With the default Wrong passwords setting, wrong passwords count separately for each network they come from, so guesses from one place cannot lock the user out everywhere, and an address that keeps failing from many networks is slowed instead. A tenant can choose to lock password sign-in for the address everywhere instead; Wrong passwords explains both. Each limit, and how many users it currently holds back, appears on your tenant's usage view.
Sign-in requirement
Sign-in requirement on the Authentication page makes every sign-in by the people it covers use a stronger method: a passkey or security key, or two different kinds of factor. It covers people who hold a management role, the default, or everyone. It has three modes.
- Off, the default.
- Report-only changes no sign-in. Each sign-in by someone it covers that finishes without the method is counted for that person and day, with the first step they used, and kept for 90 days. The Authentication page shows the last 7 days' count, and Governance, Reports names the people under Sign-in requirement, so you can help each of them set up the method before you enforce it. Report-only can run before the method is turned on.
- Enforced asks for the method as the second step, so only methods that meet it are offered: a passkey for a passkey requirement, or a method of a different kind from the first step: an authenticator app or passkey, never an email code. Recovery codes and push approvals are not offered for it, and a trusted browser does not skip it. Someone who has not set up the method signs in the usual way, including any second step they already have, then adds it before continuing, as in a forced migration. Enforcing needs the method turned on.
The requirement applies to sign-ins on your tenant's pages and at its authorization endpoint, including sign-ins through another identity provider. Someone who lost their passkey cannot get past an enforced passkey requirement with a recovery code: an administrator resets their sign-in methods in Users, and they set up a new one at their next sign-in. A sign-in that sets up the method counts as meeting it.
Sessions that already started are not signed in again by themselves. When a change asks more of people, because the requirement becomes enforced, asks for another method or covers everyone instead of role holders, the page offers Sign out existing sessions it covers that do not meet it. Checked, every session on your tenant's pages that belongs to someone the requirement now covers and did not use the method ends when you save, and those people sign in again under the requirement. Left unchecked, those sessions keep working until they expire or end: they can still approve OAuth sign-ins and device sign-ins, while tenant-hosted management waits until each person meets the requirement. Sessions of Beyond the Login accounts are never ended this way; they belong to their owner. Audit records which choice was made and how many sessions ended. The mode is part of the authentication settings, so changing it needs the Authentication update permission and is recorded in Audit; reading the names needs the Governance reports permission, and each read is recorded too. A point-in-time restore keeps a requirement enforced since the restore point.
What users see
Tenant users sign in on your tenant's own address, at /login. Applications that use your tenant send them there during an authorization request, and they return to the application once every step is done. The sign-in page asks for the email address once, with a button on the same form for the password and for each email method you turned on, and a Sign in with a passkey button when passkeys are on, plus Create an account and Forgot password? when those apply. While Remember the last email address in the Sign-in page card of Authentication is on, as it is by default, the page fills in the address of the last successful sign-in on that browser, kept in a cookie for a year, with an X at the end of the field that clears it and forgets it. It never signs anyone in or skips a step, and an application's login_hint takes its place.
Signed-in tenant users have their own Your Profile and Security pages at /account and /account/security. There they can confirm their email address, change their password, turn email sign-in on or off, set up an authenticator app or passkey, create new recovery codes, and forget trusted browsers. A user holds one authenticator app at a time, so moving to a new app or phone means removing the old one first and then setting up the new one; several passkeys can be added side by side. Changing sign-in methods needs a recent sign-in, and changing a password also needs the current one. Users receive an email whenever a method is added or removed.
New passwords follow the tenant's password rules and, unless the tenant turned the check off, are checked against passwords known from data breaches, at registration, reset, on the Security page and when an administrator sets one. If the check cannot run, the password is not accepted and the user can try again shortly. People see where they are signed in on their Security page and can sign other browsers out. A password change or reset ends the user's other sessions and signs them out of applications, and a reset also ends any sign-in links and codes still waiting in their mailbox. When users change their own email address, their other browsers and applications are signed out and links already sent to the old address stop working.
When an administrator changes a user's email address in Users, the new address must be confirmed before email sign-in and codes work again, and the user is signed out everywhere. Because password reset links go to that address, changing it also needs the permission to set user passwords. Changing only the user's name keeps them signed in; applications see the new name the next time they receive tokens.
Sign-in methods in tokens
ID tokens carry an amr claim listing how the user signed in, using the values registered in RFC 8176 where one fits. pwd means a password. otp means a one-time secret: an email sign-in link, an email code, an authenticator app code or a recovery code. A verified WebAuthn assertion adds pop, proof of possession of a key. Credential backup flags cannot establish whether a key is hardware-secured, so this service does not infer the more specific hwk or swk values from them. A push approval adds mca, multiple-channel authentication. mfa means two different kinds of factor were proved, or a passkey verified its user with a PIN or biometric. An email link or code never adds mfa, because email is not an authentication factor, and a push approval adds it only when the approving browser was itself signed in with two kinds of factor. The tenant emits only registered values. Sessions from before this change may still show the older values passkey and email, and still meet a passkey requirement or level.
A user who signs up through the registration page gets otp, whether they confirmed with the link or the code. A user who sets up a required method while signing in also gets that method's value, but not mfa, because the method was not proved independently of the first step.
Tenant users with management roles
A tenant user can manage your tenant when you give them a management role. In Users, choose Management roles on their row and select the built-in Tenant Admin role or a custom role. They then manage the tenant at /manage on your tenant's address, and can do exactly what their roles allow. A tenant user's roles never reach Beyond the Login or another tenant, and at least one BTL Tenant Admin must always remain.
BTL accounts
Beyond the Login accounts follow the same kinds of settings, managed by BTL administrators for BTL's own organization. Your account's Security page shows the methods BTL offers and the ones you have set up. The BTL organization has no sign-in requirement yet: BTL accounts sign in on the Beyond the Login site, which does not apply one, so its Authentication page leaves the setting out. Of the registration settings, only whether people can change their own email address applies there: BTL users change it on their Profile page. Accounts are created by signing up on Beyond the Login, so self-registration is not a BTL setting.
A BTL user also signs in on their own personal tenant's address, under that tenant's settings. Passkeys they add on their Beyond the Login Security page work there too, so a passkey second step or a passkey sign-in requirement in that tenant can be met with them. They are added only on Beyond the Login, work only on the person's own tenant and the site, and a passkey answer given on the tenant's address is not accepted by the site. A passkey alone cannot start a sign-in there: the person signs in another way first, then uses the passkey.