Remove a leaver's access in the right order
Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.
Disabling a departing employee’s main login can leave an API token working, a local workspace password usable or a shared credential unchanged. It can also stop a scheduled export or leave business assets without an accessible owner.
The right order depends on whether the departure is planned or immediate. For a planned departure, identify dependencies and prepare supported ownership changes before the authorised cutoff, then revoke access and verify both security and continuity. For an immediate departure, contain access first; pause dependent work if necessary rather than leaving the account usable while someone investigates.
This employee leaver workspace access checklist centres on a filled runbook you can adapt. It covers access termination and account dependencies—not employment decisions, customer handoff content or data-retention policy. Confirm which transfer, suspension and revocation controls your workspace and current plan actually support before acting.
1. Set the cutoff, authority and definition of done
A departure needs one accountable coordinator, but not one person doing every task. The coordinator maintains the record; authorised administrators execute changes; business owners confirm that essential work still functions; a verifier checks the result.
Record an exact cutoff with a time zone and identify who can authorise an earlier shutdown. Avoid instructions such as “remove access after their last meeting”: they leave administrators guessing if the meeting changes or the employee signs in elsewhere.
Use two separate completion states:
- Access contained: known entry routes have been disabled or revoked, relevant shared secrets have been addressed, and unresolved routes have an explicit containment action.
- Departure closed: business dependencies have an accountable owner or an approved pause, verification evidence is recorded, and no access questions remain unexplained.
Do not mark a task complete merely because someone clicked a control. An administrative action is evidence of an attempt; a status check, access result or successful successor operation is evidence of the outcome.
2. Inventory routes into the workspace, not just the user record
Start with the identity directory and workspace membership, then investigate routes that may exist independently. Treat each item below as something to check, not as a claim that every product supports it.
- Interactive access: central sign-in, local passwords, guest memberships, mobile sessions, remembered browser sessions and connected devices.
- Recovery and elevation: recovery addresses, recovery phone numbers, backup codes, administrator roles and delegated privileges.
- Non-interactive access: personal API tokens, application passwords, authorised third-party applications, scripts and integrations.
- Shared access: team passwords, recovery material, shared mailboxes and credentials the employee could retrieve from a vault.
- Independent systems: external file stores, reporting tools, messaging administration and vendor portals outside central sign-in.
For each route, record its owner, how it will be revoked and how that revocation can be checked. If a session-revocation control is unavailable, document the gap and the next containment step rather than assuming that disabling login ended existing sessions.
Keep credentials out of the runbook. Record a vault item reference or token identifier, not the secret itself. The runbook may need wider review than the underlying credentials should receive.
3. Build a dependency register before changing ownership
A dependency register connects access removal to the work that could break. “Owns reports” is not enough. Identify the particular asset, the identity it relies on and the person who will prove the replacement works.
Separate ownership from execution credentials. Moving a report to a team folder may change who can find it without changing the personal token used to populate it. Similarly, giving a colleague access to an automation does not establish that the automation can run without the leaver.
Fictional worked example: Every person, system label, time and result in the runbook below is hypothetical. The listed controls must be checked in the actual services before use.
Filled dependency register: Maya Chen’s departure
- DEP-01 — Weekly pipeline export. Current dependency: scheduled job authenticates with Maya’s personal token. Business owner: Priya Shah, sales operations. Replacement: a separately authorised service identity, if supported; otherwise Priya runs an approved manual export. Planned action: configure and test replacement before cutoff, then revoke the old token. Immediate action: revoke the token first and pause the schedule. Evidence required: old token identifier marked revoked; replacement output accessible to Priya; record count reconciled to the selected source view.
- DEP-02 — Shared sales dashboard. Current dependency: Maya is the only recorded owner. Successor: Omar Reed, team lead. Planned action: use the supported ownership-transfer process and confirm Omar can administer it. Immediate action: administrator restricts Maya’s access, then performs recovery or transfer. Evidence required: owner record and Omar’s successful administrative check. If transfer is unsupported, rebuild or recover through the provider’s supported process.
- DEP-03 — Messaging workspace administration. Current dependency: Maya holds an administrator role and is listed as a recovery contact. Successor: Lena Ortiz, authorised workspace administrator. Planned action: establish and test independent administration, then remove Maya’s role and recovery route at cutoff. Immediate action: Lena uses her existing independent access to contain Maya’s routes. Evidence required: current role list and recovery-contact review. If no independent administrator exists, invoke provider recovery rather than sharing Maya’s login.
- DEP-04 — Shared vendor credential. Current dependency: Maya could retrieve a shared password used by a reporting connector. Secret custodian: Lena. Consumer owner: Priya. Planned action: prepare replacement, update the connector and retire the old credential at cutoff. Immediate action: invalidate the old credential first; accept a reporting pause until the connector is updated. Evidence required: vault version reference, connector test and confirmation that the previous credential no longer authenticates through an approved check.
- DEP-05 — Private working folder containing business documents. Current dependency: Maya’s account owns the folder. Successor: Omar. Action: confirm the supported administrative transfer or recovery method before deletion. Evidence required: Omar can open and manage the required documents using his own account. Any unresolved disposition follows the organisation’s existing data policy.
4. Execute the planned-departure branch
The following filled runbook uses a hypothetical Friday cutoff of 16:00 UTC. Its timings illustrate sequence, not a prescribed notice period.
Runbook header and responsibilities
- Case: LEAVER-027; employee: Maya Chen; workspaces and identifiers linked in the restricted case record.
- Authorising manager: Omar Reed; confirms scope and cutoff.
- Coordinator: Alex Morgan, operations; maintains the route inventory, dependency register and exception log.
- Access administrator: Lena Ortiz; executes authorised identity, workspace and secret changes.
- Business dependency owner: Priya Shah; validates exports and reporting continuity.
- Verifier: Noah Bell, an authorised administrator who did not execute the changes; reviews evidence and performs permitted checks.
Filled sequence
- Thursday, 10:00 UTC — Alex opens the case. Omar confirms Friday’s cutoff and the immediate-departure authority. Alex inventories access routes and records DEP-01 through DEP-05. Evidence: scoped account list and manager authorisation reference.
- Thursday, 14:00 UTC — Lena confirms available controls. She records the supported methods for ownership changes, login suspension, session revocation, token removal and recovery-contact updates. Unsupported controls become named exceptions, not assumptions.
- Friday, 11:00 UTC — Priya and Omar validate replacements. Priya checks the replacement export against the selected source view. Omar confirms dashboard administration and document access using his own account. No successor uses Maya’s credentials.
- Friday, 15:30 UTC — Alex runs a readiness check. Every dependency has either a tested replacement or an approved pause. Lena prepares the secret change without putting the replacement secret in the case record. Unneeded parallel schedules are disabled to prevent duplicate outputs.
- Friday, 16:00 UTC — Lena applies the access cutoff. She blocks central and independent login routes, removes relevant memberships and roles, revokes supported sessions and personal credentials, and removes recovery routes. These are separate checklist entries, even when one control covers several routes.
- Immediately after cutoff — Lena retires exposed shared secrets. She updates authorised consumers and checks their operation. Where old and new credentials briefly coexist, the old credential receives an explicit retirement task and verification result.
- After execution — Noah verifies containment; Priya verifies continuity. Alex records the evidence, remaining exceptions and next checks. Account deletion is not used as a substitute for these steps.
Preparation reduces disruption, but it does not justify leaving old access active past the authorised cutoff. A dependency that fails readiness becomes a paused service with an owner—not a reason to keep the leaver’s account available.
5. Switch to containment for an immediate departure
Use the same inventory and register, but reverse the priority: close access before completing transfers. Do not wait for the departing employee to provide an inventory or approve changes.
- Omar authorises immediate containment; Alex records the time and scope. Lena works from existing administrative access, not the leaver’s credentials.
- Lena blocks known entry routes. Central login, independent local access, privileged memberships, sessions, tokens and recovery routes are addressed according to supported controls. Record each action’s time rather than claiming they happened simultaneously.
- Lena invalidates relevant shared secrets. If updating a dependent connector would delay containment, invalidate first and let Priya pause the affected workflow.
- Alex reconciles the inventory. Newly discovered routes receive an owner and an immediate containment decision. Unknown token ownership or an unavailable revocation control remains an open exception.
- Priya and Omar recover business functions. They use supported administrative transfers, replacement credentials or a documented manual process. Lena does not casually re-enable the former account to restore a job.
- Noah verifies closure of each route. Alex keeps the case open until unresolved access questions are resolved or escalated with a concrete containment measure.
For this hypothetical case, the immediate branch pauses DEP-01 and DEP-04 while preserving independent administration of the messaging workspace. That interruption is deliberate: it avoids keeping a personal token or shared password valid merely to preserve reporting.
6. Record evidence that answers a specific question
A useful completion record says what was tested, who tested it and what the result establishes. Store sensitive logs and screenshots in restricted locations; put references in the case.
Hypothetical filled completion record
- Central login — Lena executed; Noah verified. Account status shows blocked at Friday 16:00 UTC. Authentication record reviewed for subsequent successful sign-ins; none found in the reviewed interval. Evidence: restricted log reference AUTH-027. Follow-up: check again after the next relevant activity cycle.
- Workspace sessions — Lena executed. Supported session-revocation action recorded at 16:02 UTC. Noah reviewed the resulting session state. Evidence: SESSION-027. Limitation: this does not establish the state of unrelated external services.
- Personal export token — Lena revoked; Priya validated replacement. Token identifier recorded as revoked; approved test of the old credential rejected; replacement export completed against the selected source view. Evidence: TOKEN-027 and EXPORT-027.
- Dashboard and documents — Omar verified. Successor ownership recorded and administration checked under Omar’s own account. Evidence: OWNER-027. Merely opening a document would not prove ownership transfer.
- Shared vendor secret — Lena rotated; Priya tested consumer. Previous version retired, replacement connector test successful. Evidence: vault version reference and CONNECTOR-027; no password copied into the record.
- Case status — Alex recorded; Noah reviewed. Access contained, dependencies restored, no unexplained access exceptions. Final closure follows review of the next scheduled export.
Use approved test methods. Do not collect the leaver’s password or impersonate them to manufacture proof. Where a negative authentication test is unavailable, state exactly which administrative status and logs were reviewed and what uncertainty remains.
7. Recover from failures without reopening broad access
A scheduled job stops after suspension. Check which credential the job actually used. Replace it with an authorised identity where supported, or run the documented manual fallback. Restoring the entire former account may reopen unrelated access and obscure the original cause.
An ownership transfer is unsupported. Keep interactive access blocked while using the provider’s supported recovery or export process. Delay destructive changes until the asset dependency is understood. Do not assume that another administrator can recover every private asset.
A shared-secret change breaks several consumers. Use the consumer list to update each one. If the list was incomplete, pause affected jobs while discovering the missing consumer rather than silently restoring the exposed credential.
An integration continues after login is disabled. Treat that as evidence of a separate access route. Identify its credential and owner, revoke or replace it, and extend the inventory. The login action was not necessarily ineffective; it was insufficient for that route.
A device cannot be reached. Record its status and apply available server-side revocation controls. Do not describe remote removal as completed without confirmation. Account access revocation also does not prove that previously downloaded information has disappeared.
8. Close on observable outcomes, not a checked box
Before closing the next departure, require a named owner and evidence reference for every discovered access route and dependency. Report three counts: unexplained access routes, dependencies without a tested replacement or approved pause, and overdue verification tasks. Any non-zero count needs a named exception and next action.
As the next practical step, choose one upcoming planned departure and fill the dependency register before scheduling account changes. Ask a second administrator to identify any row whose completion claim cannot be independently checked. Fix those evidence gaps before cutoff, when there is still time to protect both access and business continuity.
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