Sign in

Blog · AI and self-service analytics

Shared cloud dashboards: permissions to test before inviting your team

Evaluate cloud dashboard collaboration with separate users, explicit role and data scopes, negative access tests and a record of failed or untested requirements.

The short answerTest shared-dashboard permissions with separate identities, agreed data scopes and both allowed and denied actions. Confirm viewing, editing, exporting and administration independently, including behavior after access changes. A successful sign-in or hidden menu does not establish that the underlying data and actions are restricted as required.

Test shared cloud dashboard permissions using separate people and explicit expected results. Define who may view which records, who may change a report, who may export its data and who administers access. Check the denied actions as carefully as the allowed ones. Seeing a sign-in screen or a role name is not enough to establish the access behavior your team requires.

The metric-governance guide explains governed definitions and analytical scope. This article owns evaluating collaboration permissions in a proposed service.

Distinguish identity, role and data scope

Identity answers who is signed in. A role describes permitted actions. Data scope identifies the records those actions may apply to. Keep the three separate in the evaluation brief, because an editor for one region need not be an editor for every region.

OWASP's authorization guidance distinguishes authorization from authentication and recommends checking permissions for every request, with denied access as the default when permission is absent. These are technical review principles, not evidence that a particular product implements them.

Ask the supplier how the proposed arrangement enforces the requirement, then request behavior that a reviewer can inspect. A menu that disappears is a presentation observation; it is not by itself a test of the underlying action.

Write an expected-access matrix

List named test identities, required data and actions before opening the product. Use synthetic records whose ownership is obvious. Do not use live customer details merely to test whether an unauthorized user can see them.

Test identity Permitted records Permitted actions Expected denied action
DEMO-WEST-OWNER West View and edit agreed report View East-only account detail
DEMO-EAST-REVIEWER East View and export agreed scope Change the revenue definition
DEMO-ADMIN Approved business scope Administer agreed access Access an unrelated business

The matrix is an invented requirement, not a description of Covirage's current roles. Confirm whether the proposed service can supply these behaviors before treating them as available.

Record a synthetic test result

DEMO-ACCESS-01 runs four agreed tests for each of three identities: twelve tests in total. Nine pass, two fail and one is blocked because the required environment is unavailable. The passed proportion is 9 ÷ 12 = 75%.

Result Count Purchasing meaning
Passed 9 Evidence for those specific tests
Failed 2 Observed behavior contradicts the requirement
Blocked 1 No conclusion until the dependency is resolved

Do not remove the blocked test from the denominator to present a stronger result. More importantly, a mandatory unauthorized-access test that fails can block approval regardless of the percentage. Counts summarize the review; they do not replace its critical findings.

Test beyond the first chart

Where the proposed product supports them, include detail views, downloaded data, search results and analytical answers. A scoped chart is insufficient if another supported route reveals wider records. Test the same requirement in separate sessions rather than switching labels inside one demonstration.

Ask what happens after a role or account assignment changes. Agree how quickly the change must apply, and inspect the relevant behavior. Removal of application access does not automatically retrieve files a user already exported; external dissemination needs its own review.

Record the expected and observed result for each route or action tested. Where supported, include a direct detail link as well as normal navigation. Preserve the test identity and scope so a reviewer can distinguish a genuine denied-action check from a demonstration performed with administrator access.

The external report-sharing guide covers that later boundary, including recipients who do not have dashboard access.

Keep deployment and permissions separate

A shared dataset needs an actual shared storage and access arrangement. The deployment guide explains why opening a web application or signing into a business account does not establish that arrangement.

Record the environment, plan, assistance and data used for the tests. Preserve failures and untested paths. Ask for separate evidence when a different environment, user population or permission model is proposed. Do not turn a narrow checklist into a claim about every security control.

Bring the matrix to a scoped discussion

Inspect the synthetic customer-growth review and identify which audiences need which output. Then contact Covirage with the proposed access matrix and available inputs. Agree the review scope and the permission evidence required for those audiences.

Questions people ask

Is sign-in the same as permission to see all business data?

No. Identity and authorization are different questions. Specify which records and actions each person is permitted to access.

Should an access test include exports?

Yes, when exports are part of the proposed service. Verify that exported data obeys the required scope instead of checking only the visible chart.

Does a passed permissions checklist prove security certification?

No. It records the behavior and scope tested. Certification, wider security controls and other environments require separate evidence.