Back to all articlesAccess and workspace governance

Replace shared staff logins without disrupting daily work

Alactic Editorial9 min read

Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.

A staff member leaves, but changing the shared CRM password could interrupt everyone still working. Nobody knows whether that login also owns scheduled work, receives recovery codes, or keeps a reporting connector connected.

For a planned transition, map those dependencies, confirm available individual access, test real tasks, and retire the shared credential in stages. Where replacement is blocked, document a bounded exception rather than letting sharing remain an invisible default.

Urgent access risks take priority: if a departed person retains access, credentials may be exposed, or recovery is compromised, escalate and contain access promptly using supported controls. Do not wait for a complete inventory. Accept a controlled interruption or supervised fallback when necessary; assess dependent-service disruption alongside containment, not instead of it.

The goal is individually attributable access that lets you remove one person's access without stopping everyone else's work. This concerns staff credentials, not shared customer phone numbers or inbox assignment.

1. Separate the identity from its dependencies

A shared login can bundle human access, record ownership, recovery, and service authentication. Replacing the sign-in does not necessarily replace those dependencies.

Walk through an ordinary shift. Ask staff which identity they use, how they authenticate, and what happens when sign-in fails. Do not put passwords, tokens, or recovery codes in the worksheet. Record:

  • Human access: current and former users, devices, credential-sharing routes, and required tasks.
  • Work ownership: assigned records, private views, scheduled reports, or other work associated with the identity.
  • Recovery: who controls recovery addresses, authentication devices, and emergency material.
  • Connected services: whether each service uses a password, user session, token, or another supported method.
  • Administration: who can change access and restore service, where those controls exist.

Treat “unknown” as an investigation task. A connector that currently works is not evidence that it will survive credential rotation. Equally, an unknown dependency is not a reason to leave a departed person's access intact.

2. Confirm the destination before changing access

Check the actual workspace and current plan for individual accounts, available permissions, ownership transfer, and revocation controls. Do not assume equivalent features exist across products or devices.

Match access to tasks. Someone updating customer records may not need exports, billing, or user administration. This follows the least-privilege approach in OWASP's authorization guidance; it is a design recommendation, not evidence of a particular product's controls.

Choose a practical destination: individual access for all routine work, individual access with a narrow legacy exception, or a documented blocker requiring a different workflow, plan, or system. Compare available seat costs with the supervision, disruption, and uncertain attribution of continued sharing. Do not compensate for missing permissions by making everyone an administrator.

The following filled worksheets form one reusable transition record. Replace their entries with your own findings, keeping unknowns and failed tests visible.

3. Worksheet A: map one account before piloting

Fictional worked example: all people, quantities, dates, test outcomes, and decisions below are hypothetical. They illustrate a recommended workflow, not Alactic capabilities or customer results. The example assumes the stated account options have been verified in the fictional workspace. It begins as a planned transition, with no known departure or exposure requiring urgent containment.

  • Identity and scope: Front Desk Legacy; six current staff use it for record updates and follow-up work.
  • Owner: Maya, operations lead. Technical owner and backup coordinator: Rowan, workspace administrator.
  • Destination: six individual accounts are confirmed available. Tested staff permissions permit ordinary updates without user administration; the budget owner has reviewed seat costs.
  • Ownership: some records belong to Front Desk Legacy. Test reassignment on designated records before moving the wider workload.
  • Saved work: a private follow-up view needs its criteria documented and reproduced for individual users.
  • Scheduled work: weekly summary ownership and recipients remain unverified. Wider cutover is blocked until continuity is tested.
  • Connected service: reporting connector authentication is unknown. Rowan must identify its method, credential owner, supported replacement process, and revocation implications before planned credential changes. An urgent exposure would override that sequence.
  • Recovery: currently tied to a current shift lead's phone. Establish and test an organisation-controlled route using supported settings before routine cutover. Compromise or departure would trigger prompt containment instead.
  • Devices: reception computer and assigned laptops. Test sign-out and next-user sign-in on the reception computer.
  • Previous credential holders: six current staff are known; Rowan must check for earlier recipients and saved copies. Removing vault access will not invalidate passwords already known.
  • Historical limitation: earlier actions under Front Desk Legacy remain ambiguous without independent evidence.

This inventory separates two questions: can work move, and can old access actually be removed? Both need evidence before describing the legacy identity as restricted.

4. Worksheet B: test tasks and access removal

Successful sign-in is only the beginning. Use designated test records, avoid customer sends or destructive changes, and test both required work and prohibited actions.

Hypothetical pilot: Lena and Omar test individual access before the other four staff move.

  1. Authentication: both sign in independently. Recovery responsibilities are documented without sharing recovery secrets. Result: pass.
  2. Visibility: both open designated records and the follow-up list. Omar's list lacks one filter. Decision: correct it and repeat the test before expansion.
  3. Updates and attribution: both save an approved test-field change. Available activity history shows their respective identities. Result: pass for this action only, not proof of equivalent logging for every action.
  4. Restricted access: neither can open user administration. No settings are changed during the test. Result: pass.
  5. Ownership: a designated record moves from the legacy identity to its intended owner, remains visible in the expected list, and can be updated. Result: pass.
  6. Device handoff: Lena signs out and Omar signs in. The displayed identity is Omar's. Result: pass; staff must still check the active identity before editing.
  7. Removal: a designated test account loses new sign-in access, but an existing session remains usable. Result: fail for complete removal. Rowan must verify supported session revocation and other access paths.

Record each result with its scope. A failed removal test should change the departure procedure, not disappear beneath an overall “pilot passed” label. Password changes, session termination, token revocation, and recovery changes may be separate controls.

5. Worksheet C: stage cutover with explicit gates

Use a lower-pressure work period with an authorised administrator available. These gates describe decisions, not automatic execution.

  • Gate 1 — pilot: accounts, test records, recovery checks, and a blocker-reporting route are ready. Hypothetical status: passed.
  • Gate 2 — expansion: correct Omar's filter; verify weekly summary continuity; document a tested response to the session-removal failure. Hypothetical status: pending. Maya cannot yet approve expansion.
  • Gate 3 — routine cutover: after Gate 2 passes, move the remaining staff at an agreed shift boundary. Maya checks each person's first completed task and its available attribution.
  • Gate 4 — legacy restriction: first identify connector authentication and plan its continuity or controlled interruption. Safely rotate or otherwise invalidate previously shared credentials; check and apply supported session, token, recovery, and other relevant access changes. Verify former credential holders cannot use the tested routes. Only then describe credential custody as restricted to named custodians.
  • Gate 4 blocked: if old credentials cannot yet be invalidated, removing access to the credential store is insufficient. Record continued access by former holders as unresolved exposure. Maya must explicitly decide whether to pause the dependent work or accept a bounded exception; no one should label custody restricted.
  • Gate 5 — retirement: confirm the connector and summary work without the old human login, ownership is resolved, and recovery is controlled. Assess disabling or retaining the identity for a documented non-routine purpose.

Before disabling or deleting the identity, check effects on history, records, services, and recovery. Deletion is not a substitute for verifying removal of the access paths that matter.

6. Worksheet D: give the exception an expiry

A temporary exception cannot make a shared identity fully attributable. Nor can a restricted credential store undo previous disclosure. This filled record preserves that distinction.

  • Exception: retain Front Desk Legacy while the reporting connector's authentication dependency is resolved.
  • Business reason: an unplanned disconnect could interrupt the weekly operations report.
  • Owner and permitted work: Rowan investigates the connector during an approved maintenance period. No routine customer-record work is authorised under the legacy identity.
  • Start and expiry: hypothetical start, 5 October 2026; expires 12 October 2026 at 17:00 UTC.
  • Renewal approver: Maya. Before expiry, she must approve retirement, pause the dependent activity, or explicitly renew with a new expiry, reason, unresolved exposures, and next action. No silent rollover.
  • Current exposure: the credential store is limited to Rowan and designated backup Priya, but previously shared credentials have not yet been invalidated. Former holders may still have access. Restricted custody is not established.
  • Required containment work: identify connector authentication; plan supported credential invalidation and session, token, and recovery changes; test service continuity and access removal. If continuity cannot be preserved, document the controlled interruption or supervised fallback.
  • Operator record: record operator, purpose, start and end times, and affected configuration. One operator at a time where practical. This supports investigation but does not prove who performed every application action.
  • Progress review: Maya checks progress on 8 October 2026. This does not replace the expiry decision.
  • Early containment trigger: suspected exposure, loss of recovery control, or a custodian's departure requires prompt assessment and supported credential, session, token, and recovery changes as appropriate. Consider dependent-service disruption, but accept controlled interruption when needed rather than waiting for the inventory.
  • Exit evidence: supported connector authentication is established, the report is tested, old credentials are invalidated, relevant revocation checks pass, and no remaining human task requires the shared identity.

If restrictions cannot be enforced, say so. Naming custodians and adding an expiry documents accountability; it does not itself remove access.

7. Recover narrowly and verify the next shift

Rollback should address the failed task, not reopen shared access to everyone. If records disappear, pause that task and inspect visibility and ownership rather than granting administrator rights. If the connector stops, check authentication and whether any work completed before rerunning it, to avoid duplicate output.

If someone cannot authenticate, use approved recovery or supervised coverage, not another employee's credentials. If access removal is incomplete, investigate supported sessions, tokens, recovery routes, and other entry paths; record unresolved exposure and escalate. Any essential temporary legacy operation needs a named operator, limited task, explicit authorisation, and end time.

At the next shift review, count staff who completed required tasks individually, routine tasks still requiring the shared login, unresolved dependencies, and tested access-removal routes. Keep attribution coverage separate: checking record updates does not establish visibility into every action.

For this hypothetical team, Rowan's next tasks are session-revocation verification and connector authentication discovery. Maya can then decide whether expansion is justified. The meaningful outcome is not six new accounts; it is routine work continuing under identifiable users while old access is demonstrably removed or honestly recorded as unresolved.

Access governanceShared accountsCRM operationsStaff transitions

Put the next conversation to work.

Explore Alactic Communications for WhatsApp conversations, campaigns and AI-assisted workflows. Choose the product and plan that fit your team.

Explore Communications