Keep mobile access within the same workspace boundaries as desktop access
Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.
Removing someone from a workspace on desktop is an administrative action, not proof that every access path is closed. An Android app might still show an already-open record. A mobile browser might hold a different signed-in account. A saved link might lead straight to a workspace object without passing through the workspace selector.
To verify mobile and desktop workspace access permissions, inventory the entry points people actually use, define the expected boundary for each identity, and test that boundary on every supported surface. Repeat the relevant checks after a role change or removal. Record what the user could read or do—not just whether a button disappeared.
The goal is consistent authorisation, not identical interfaces. A feature absent from an app is not automatically a permissions success, and different sign-in behaviour is not automatically a failure. The useful question is whether each supported route respects the same business access decision.
1. Define the boundary before testing the screens
Start with a short access statement that a reviewer can evaluate. For example: Hypothetical policy: a restricted operator may read assigned records in Workspace Cedar, cannot change membership, and has no access to Workspace Birch. Name the permitted data and actions rather than relying solely on a role label.
Separate three decisions:
- Workspace membership: which workspace the identity may enter.
- Object access: which records or conversations that identity may read within the workspace.
- Action permission: what the identity may change, send, export, assign, or administer, where those functions exist.
A workspace selector can appear correct while a saved record link still exposes information. Similarly, hiding an administrative control does not establish that the action is rejected when attempted through another supported route.
OWASP's authorisation guidance recommends least privilege, denying access by default, and checking permissions on every request. Those principles explain why verification should go beyond visible navigation; they are not evidence that a particular product implements a control.
For Alactic, check the relevant product, workspace, current plan, and supported surfaces before selecting tests. Do not assume that CRM functions also exist in Communications, that ACID has the same permissions model, or that every web feature is available in the Android app.
2. Inventory entry points, identities, and session states
Use actual working routes rather than a generic device list. Desktop web, mobile web, and an Android app are separate test surfaces when supported and used. Within each, distinguish a fresh sign-in from an already-open session.
For each route, record:
- The product, workspace, browser or app, and version where available.
- The test identity, expected role, and permitted workspaces.
- The entry path: ordinary navigation, workspace selector, saved link, or app link where supported.
- The session state: fresh sign-in, active session, backgrounded app, or previously selected account.
- The test object and expected read or action boundary.
Include account switching if people use more than one identity. A personal account and a work account on the same phone can make a test appear inconsistent when the tester has actually changed identities. Record the signed-in identity explicitly instead of inferring it from an avatar or the last workspace opened.
Use authorised test identities and synthetic data. Choose harmless actions with reversible outcomes. Where testing a real send, export, or destructive change would create risk, use an approved test environment or mark the case untested until a safe method is agreed. Do not silently treat an untested action as a pass.
3. Use a filled checklist, not an empty sign-off box
Fictional worked example: The following checklist describes an imagined internal test, not an Alactic customer result or a claim about available product controls. All identities, workspaces, outcomes, and timing are hypothetical. Substitute supported functions and your own expected policy before using it.
Hypothetical setup: Reviewer Mara checks desktop web, mobile web, and an Android app in a test workspace. Operator Sam should access Cedar but not Birch. Synthetic record Cedar-Test belongs to Cedar; Birch-Test belongs to Birch. The planned role downgrade removes Sam's membership-administration permission. A later removal ends Sam's Cedar access while retaining an unrelated permitted workspace. Evidence references below are fictional redacted capture labels.
Workspace selection and direct entry
- Case A — ordinary workspace selection. Surfaces: desktop web, mobile web, Android. Expected: Sam can select Cedar, but cannot enter Birch. Observed result: Cedar opened on all three surfaces; Birch was unavailable through the selector. Status: pass for selector behaviour only. Evidence: A-desktop, A-mobile-web, A-android. Reviewer: Mara.
- Case B — saved link to an excluded workspace record. Surfaces: desktop web and mobile web; Android link opening tested separately. Expected: Birch-Test content remains inaccessible. Observed result: both browsers denied access without displaying record content. The Android link opened in the mobile browser, so no native-app route was established. Status: browser routes passed; native deep-link route not applicable to this tested link. Follow-up: document the actual link destination.
Restricted actions and account switching
- Case C — membership change after role downgrade. Surfaces: existing desktop session, fresh mobile-web session, Android app. Expected: Sam cannot change membership. Observed result: desktop rejected the attempted change; mobile web offered no membership control, and the previously saved supported settings route denied access. The Android app had no equivalent supported function in this hypothetical setup. Status: web checks passed; Android action not applicable, not an authorisation pass.
- Case D — switch between two authorised test identities. Surfaces: desktop web and Android, where switching is supported in this example. Expected: each identity shows only its own permitted workspace content. Observed result: desktop changed cleanly. Android initially retained Cedar's header after switching to a second identity without Cedar access; opening Cedar-Test was denied. Status: follow-up required for stale workspace context; no successful record read observed. Evidence: D-android.
Access after workspace removal
- Case E — already-open record after removal. Surfaces: desktop web, mobile web, foreground Android app. Expected: no new Cedar data or successful Cedar actions after removal. Previously displayed content must be assessed separately. Observed result: refresh and new record navigation were denied on all tested routes. Android retained the previously displayed synthetic record text until navigation. Status: new retrieval checks passed; retained display requires separate risk review.
- Case F — background, resume, and reconnect. Surface: Android. Expected: returning to the app or reconnecting must not restore Cedar access. Observed result: after removal, the offline screen retained old text. Reconnecting and requesting another record produced a denial; no new content appeared. Status: online retrieval check passed; offline retained-content question remains open.
- Case G — fresh sign-in after removal. Surfaces: desktop web, mobile web, Android. Expected: Cedar remains inaccessible; the unrelated permitted workspace remains usable. Observed result: fresh sessions could not enter Cedar and could open the permitted workspace. Status: pass. Evidence: G-all-surfaces.
- Case H — switch away and back after removal. Surface: Android; repeat elsewhere if supported. Expected: returning to Sam must not restore Cedar membership. Observed result: Cedar was absent from selection, and the saved Cedar-Test route was denied. Status: pass for tested route.
Hypothetical completion record: Mara recorded the administrative change time, each subsequent attempt time, test versions, and redacted evidence. The stale header and retained text remained open findings with named owners. The exercise was not marked fully complete merely because fresh sign-in checks passed.
4. Interpret missing features and stale screens correctly
Use distinct result labels: pass, fail, not applicable, untested, and follow-up required. Add the scope of each conclusion. “Pass: fresh retrieval denied” is more useful than “mobile secure.”
A missing mobile control means that particular action may not be available through that interface. It does not show that another supported route enforces the permission. Conversely, do not invent a native-app test for a feature that exists only on the web. Record the difference and test the available route.
Stale content also needs careful classification. Separate:
- Previously rendered content: information already on screen before the change.
- New retrieval: information requested after the change.
- New action: a change attempted after the change in authority.
These outcomes carry different implications. A retained screen can still expose sensitive information to someone holding the device, even if new retrieval is blocked. But seeing old text alone does not establish that the account can still fetch current records. Record both facts without overstating either.
5. Make role changes and removals separate verification events
A downgrade should test the permissions that changed while confirming necessary access remains. Removal should test the loss of workspace access, including routes that were already open. Removing one workspace membership is not necessarily the same operation as disabling an identity across all workspaces.
Use this sequence:
- Establish the baseline. Confirm the authorised identity can perform the relevant action before the change, using synthetic data.
- Record the administrative change. Capture who made it, its scope, and the time.
- Test existing sessions first. Do not sign out everywhere before checking them; doing so would erase the condition you need to examine.
- Test fresh sessions separately. Verify the workspace and direct routes after a new sign-in.
- Check retained permissions. Ensure unrelated authorised work remains available where intended.
- Close with evidence. Record results, unresolved routes, owners, and the next verification step.
Do not invent an acceptable delay. Check documented behaviour or obtain clarification from the product administrator or provider. Record actual observation times and mark uncertainty explicitly if the expected enforcement timing is unknown.
6. Recover from a failed check without losing the evidence
If a removed identity can retrieve new restricted data or complete a prohibited action, stop using real information for further tests. Preserve a minimal redacted record of the identity, workspace, route, session state, change time, attempt time, and outcome.
Use available, authorised containment controls and follow the organisation's escalation process. Depending on supported capabilities, that may involve broader account suspension or session revocation. Treat those as actions to verify, not assumed features. Do not conclude that forcing one app to close revokes every session.
Investigate whether the issue concerns membership scope, the wrong identity, a stale interface, or successful unauthorised access. A record identifier or saved URL is not itself permission to read the record; OWASP's object-level authorisation guidance explains why access must be checked for the requested object.
After remediation, repeat the failed route in the original session conditions and a fresh session. Keep the initial observation alongside the retest. A workaround that changes the route is not proof the original problem is resolved.
7. Keep the verification workload proportionate
Testing every possible combination after every change can become impractical. Maintain a small baseline covering each supported surface, workspace exclusion, a meaningful restricted action, account switching where used, and removal from an active session. Add deeper checks for sensitive roles, newly supported surfaces, or previously failed routes.
This trades exhaustive coverage for repeatable evidence. Make the omissions visible: an unused browser can be out of scope; an untested app version cannot be described as verified. Revisit the baseline when supported access paths or permissions change.
Next action: assign an owner to inventory your current entry points and run the checklist with synthetic data for one role downgrade and one workspace removal. Report supported routes tested, expected denials observed, unexpected successful reads or actions, and unresolved cases. Completion means every in-scope route has an observed result or a named follow-up—not simply that desktop access looks correct.
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