Deploying passkeys is more than enabling a new sign-in button. This runbook helps developers and IT administrators plan enrollment, assess device support, design recovery paths, test account restoration, and manage passkey credentials throughout their lifecycle without weakening identity security.
Overview
Passkeys use public-key authentication through WebAuthn and related platform capabilities. During registration, a device or authenticator creates a key pair. The private key remains protected by the authenticator, while the service stores the corresponding public key. At sign-in, the authenticator proves possession of the private key without sending a reusable password to the service.
That model can reduce password reuse and phishing exposure, but deployment still creates operational questions. Which browsers and operating systems are in scope? Can users enroll more than one authenticator? What happens when a phone is lost, a hardware security key is damaged, or an employee changes devices? Which recovery method is strong enough for the account's risk level?
Treat a passkey rollout as an identity lifecycle project with four stages: prepare, enroll, recover, and review. Do not measure success only by the number of registered credentials. A usable deployment must also provide a tested recovery route, clear administrator procedures, and a way to remove credentials that are no longer trusted.
Core deployment principles
- Preserve account ownership. Define how the service confirms that a person is entitled to enroll or recover an account.
- Offer deliberate redundancy. Where the risk model permits it, allow users to register an additional authenticator rather than relying on one device.
- Separate authentication from recovery. A fallback process should not quietly become a weaker, permanent sign-in method.
- Test the unhappy paths. Lost devices, replaced phones, unavailable security keys, interrupted registration, and account migration should be part of acceptance testing.
- Keep the interface understandable. Users need to know what they are registering and how to identify or remove an old credential later.
Checklist by scenario
Scenario 1: Preparing a new service
- Document the accounts and actions that passkeys will protect. Mark high-impact actions such as changing recovery details, adding administrators, exporting data, or deleting an account.
- Confirm that the identity provider, authentication library, or managed service supports the WebAuthn flows you require, including registration, authentication, credential deletion, and recovery administration.
- Choose the account-binding rules. Decide whether one person may register multiple passkeys, whether administrators may revoke a credential, and how each credential will be labeled.
- Define acceptable authenticator categories for your environment. These may include platform authenticators, roaming hardware security keys, or both. Avoid assuming that every user has the same device type.
- Write the recovery policy before enabling enrollment. Specify who can approve recovery, what evidence is required, how the event is logged, and whether sensitive actions receive an additional review period.
- Prepare test accounts that represent normal users, administrators, shared operational roles, and users with restricted devices.
Scenario 2: Enrolling an individual user
- Verify the user through an existing trusted session or an approved administrative process before starting enrollment.
- Explain that a passkey is tied to an authenticator or credential-management system, not simply to a browser tab.
- Ask the user to register a primary authenticator and, where policy allows, a second independent option.
- Require the user to sign out and complete a fresh sign-in test. A successful registration screen alone is not sufficient.
- Have the user locate the credential-management page and confirm that the new credential has a recognizable name or creation date.
- Record only the operational metadata needed for support. Never ask users to disclose private key material, device unlock codes, or biometric data.
Scenario 3: Supporting a mixed-device workforce
Build a simple device-support matrix before rollout. The exact result depends on your provider, browser, operating system, device-management controls, and account policy, so validate each combination in a test environment.
| Environment | Questions to test | Recommended preparation |
|---|---|---|
| Managed desktop or laptop | Can users register and use the platform authenticator under device policy? | Test browser settings, account restrictions, and replacement-device procedures. |
| Mobile device | Can the user authenticate after screen lock, account, or device changes? | Document migration expectations and require a second recovery option where appropriate. |
| Shared or kiosk workstation | Is local credential registration appropriate for a shared device? | Prefer an individual authenticator or approved security key; avoid shared secrets. |
| Hardware security key | Can users enroll, name, test, replace, and revoke the key? | Define custody, spare-key, inventory, and loss-reporting procedures. |
| Restricted or legacy environment | Does the required browser and operating system support the intended flow? | Use a time-bounded fallback while the environment is remediated. |
Scenario 4: Handling account recovery
Use a recovery decision table rather than a single universal fallback:
| Situation | Preferred response | Control to add |
|---|---|---|
| User has another registered passkey | Authenticate with the remaining credential and register a replacement. | Review and remove the lost credential. |
| User has a managed device replacement | Use the approved migration or re-enrollment process. | Verify device ownership and record the event. |
| All authenticators are unavailable | Use the documented high-assurance recovery process. | Require identity verification, approval, logging, and notification. |
| Administrator account is locked out | Use a separately controlled break-glass procedure. | Limit access, monitor use, and test the procedure regularly. |
What to double-check
Before general release, run an enrollment and recovery test for every important account type. Capture the result in a short test record containing the account type, device and browser combination, registration outcome, sign-in outcome, recovery path, administrator involved, and unresolved issue.
Enrollment checklist
- The account is already protected by a trusted sign-in or approved provisioning process.
- The registration ceremony shows the correct service and account context.
- The user can cancel and restart without creating confusing duplicate records.
- The resulting credential appears in the account's authenticator list.
- The user can complete a new sign-in after closing the original session.
- Credential names do not expose unnecessary personal or device information.
Recovery test template
- State the failure: lost phone, unavailable key, replaced device, or inaccessible platform account.
- Identify the evidence required to establish account ownership.
- Run the recovery process with a test account, not a production account, when possible.
- Confirm that recovery does not disclose existing credential details or bypass administrator controls.
- Register the replacement authenticator and complete a fresh sign-in.
- Revoke the unavailable credential and verify that it no longer works.
- Check audit records, user notifications, and support documentation.
Also review account-change protections. A user who can add a new passkey, change the recovery address, disable security controls, and remove administrators in one session may have more power than your threat model intended. Apply step-up authentication, approval, or notification controls to sensitive changes where appropriate.
Common mistakes
- Enabling enrollment without a recovery plan. This shifts the problem from passwords to support tickets and may encourage unsafe manual overrides.
- Allowing only one authenticator. A single device creates a fragile dependency. Consider redundancy for accounts whose loss would disrupt work.
- Treating every fallback as equivalent. Email links, help-desk approval, backup codes, and another passkey have different risks. Document their intended scope and expiry.
- Ignoring credential lifecycle management. Users need a clear way to identify old, duplicated, or lost credentials and administrators need revocation procedures.
- Testing only the happy path. A rollout can appear successful until a user changes phones or an administrator loses access. Include failure and interruption tests.
- Overpromising compatibility. Support depends on the complete environment. Maintain an internal matrix instead of relying on a general statement that passkeys are supported.
- Using recovery as routine authentication. Recovery should restore access under controlled conditions, not become the easiest way to sign in every day.
When to revisit
Review this runbook before seasonal planning cycles, major device refreshes, identity-provider changes, browser or operating-system changes, and any security incident involving account takeover or impersonation. Revisit it whenever workflows or tools change, especially when you introduce a new authenticator type, alter administrator roles, or change the account recovery vendor.
At each review, sample active accounts and check for inactive, duplicate, or unrecognized credentials. Ask support teams which recovery cases caused confusion. Re-run the device-support matrix for the environments that changed, then update enrollment instructions and escalation paths.
For a practical next step, schedule a 30-minute recovery drill with one standard user account and one privileged account. Record every screen, approval, notification, and audit event. Fix the largest ambiguity before expanding enrollment. For broader identity continuity planning, compare this process with the digital persona recovery and account migration runbook and the employee passkey rollout checklist. A passkey deployment is ready when users can enroll confidently, administrators can revoke and replace credentials, and the organization can recover access without creating a weaker identity system.