Executive summary
A marketing agency can produce a polished dashboard in a day and still leave its client with a fragile reporting system. The dashboard may be owned by an employee who eventually leaves, connected through an agency-managed advertising login, shared with a broad mailing list and copied from a template that still points to another client’s data. None of these failures is primarily a chart-design problem. They are failures of ownership, credentials, access boundaries and operating procedure.
The right governance model separates four decisions: who owns the report, who can edit or view it, whose credentials authorize queries, and which underlying records the query can return. These decisions interact, but they are not interchangeable. A client may own a report without owning its connector authorization. A user may have permission to open a report but not permission to query a viewer-credential data source. Conversely, a report using owner’s credentials can display data to people who do not have direct access to the underlying system.
For agencies, the practical answer is not to prohibit templates or insist on an expensive enterprise platform. It is to establish client-specific data boundaries, a documented access matrix, controlled report creation, verified handover and a lightweight audit trail. Use the framework below to make reporting reusable without making client data interchangeable.
1. Define the assets before deciding who owns them
An agency reporting engagement usually includes more assets than the visible dashboard: advertising accounts, analytics properties, CRM connections, spreadsheets, API connectors, warehouse datasets, service accounts or OAuth authorizations, data sources, calculated fields, report templates, delivery schedules and documentation. A dashboard URL does not transfer ownership of this entire chain.
Create a client asset register with one row per production report and its dependencies. Minimum fields are client identifier, business owner, report owner, technical maintainer, report URL, environment, data source IDs, source system, credential mode, credential administrator, authorized audiences, last access review, last reconciliation, refresh expectation and exit plan. Treat the register as an operational record, not a compliance document written once and forgotten.

Assign four distinct accountabilities. The client business owner approves definitions and intended audience. The agency delivery owner maintains the report and coordinates changes. The data owner authorizes access to source systems. The platform or identity administrator controls accounts, roles and credential lifecycle. One person can hold more than one role in a small engagement, but the responsibilities should still be explicit.
2. Understand the three permission layers
Layer A: access to the report
Report permissions decide who can open, edit, share or administer a report. Viewer and editor access are not the same as permission to see every row in an upstream database. Grant named users or controlled groups the minimum required role. Reserve editor access for people who genuinely need to change structure, fields, filters or sharing. A client executive usually needs viewer access, not editor access.
Layer B: access to the data source
Data sources are independent assets with their own schema, connection settings and sharing behavior. Reusable data sources can be managed separately from reports; embedded data sources live with the report and are more readily changed by its editors. Review both report access and data-source access, especially when staff can create copies, edit calculated fields or replace connections.
Layer C: credentials that execute the query
Google’s data credentials documentation distinguishes owner’s credentials, viewer’s credentials and service-account credentials. Owner’s credentials allow a report to retrieve data under the credential owner’s authority, even when viewers do not have direct source access. Viewer’s credentials require each viewer to be authorized for the underlying data. Service-account credentials can remove dependence on a named individual in supported configurations, but support and setup are connector-specific. Do not assume every marketing API connector supports service accounts.
These three layers should be tested independently. Granting someone access to the report does not prove that they should see its underlying data. Likewise, removing a person from a source system may not stop them viewing a report that still uses a different owner’s valid credentials. See Google’s explanation of data credentials for the authoritative distinctions.

3. Client isolation must be enforced at the data boundary
The most important agency rule is simple: one client’s users must never be able to retrieve another client’s records, even if a report is copied, a filter is removed, a chart is edited or a link is forwarded. A report-level filter is useful for presentation, but it should not be the only protection between clients.
Pattern 1: separate client source accounts and data sources
For a modest agency portfolio, the cleanest default is often one client-specific source authorization and one or more client-specific data sources. A client dashboard for Brand A connects only to Brand A’s advertising, analytics and CRM resources. Brand B has a separate connection and separate authorized users. This increases the number of assets but reduces the blast radius of a mistaken template or credential change.

Pattern 2: separate warehouse datasets or views
When data is staged centrally, use client-specific datasets, projects, schemas or security-enforced views as appropriate to the warehouse and contractual model. The reporting identity for Client A should not be able to query Client B’s tables. Where shared infrastructure is unavoidable, apply upstream row-level policies and test them with real client identities. A WHERE client_id = ‘A’ clause inside a chart is not a security boundary if editors can remove it.
A genuinely multi-tenant reporting model can be appropriate when client counts are high and the organization has mature identity mapping, tested row-level restrictions and controlled data distribution. It also carries more operational risk. Viewer-email filtering and related features may help in specific designs, but they are not a substitute for checking how connector credentials, exports, report editing, copied reports and scheduled delivery behave. Treat this pattern as an architecture project with explicit threat modeling and negative tests, not a quick way to reduce the number of dashboards.
For most small and mid-sized agencies, a per-client boundary with a shared visual template is easier to reason about than a shared cross-client data source with a presentation filter. Reuse layout and definitions freely; reuse data access only after proving isolation.
4. Choose embedded and reusable data sources deliberately
An embedded data source is convenient for a self-contained report: it travels with the report when shared or copied, subject to credential behavior. That convenience can be risky in an agency workflow because report editors can also modify the embedded source. A reusable data source is a separately managed asset that can support consistent field definitions across multiple reports. Its sharing and copying behavior differs from embedded sources.
Use embedded sources for contained, low-risk reports when the editor group is trusted and handover is straightforward. Prefer client-specific reusable sources for recurring executive reporting where several dashboards need the same governed metric definitions, connection settings and source ownership. Reusable does not mean cross-client: create reusable sources within a client’s boundary, not a universal agency data source pointing to every customer.
Google’s guidance on embedded versus reusable data sources also explains an easy-to-miss copy behavior: copying a report does not automatically copy its reusable data sources. Template deployment therefore needs a connection checklist, not merely a Duplicate button. Test a copied report before it is shared with a client.
5. Make templates portable without carrying credentials
A good agency template standardizes page structure, branding, date controls, metric naming, QA tests and accessibility. It does not embed another client’s authorization, hard-code a confidential customer identifier or expose production data in a demonstration screenshot. Maintain a master template using synthetic or non-sensitive sample data where practical.
Use a controlled deployment sequence. First copy the master into a client-specific workspace or ownership location. Second connect the copy to that client’s approved data sources. Third confirm field mappings, timezone, currency, attribution windows and data grain. Fourth validate totals against the client’s source platform. Fifth inspect every page, hidden page, control, blend and scheduled delivery. Sixth test access with a client viewer and an unauthorized account. Only then distribute the production link.

For example, a template may contain Spend, Leads and Cost per Lead. One advertising platform defines leads as platform conversions; a CRM defines them as qualified records. The template should not silently treat these as interchangeable. Document which source wins, whether leads are deduplicated, what timezone applies and whether the measure is a snapshot or a restatable historical value.
6. Use a practical agency access matrix
Agency delivery team
Grant limited report editor access to named maintainers who need to implement changes. Restrict their ability to expand audiences where available, and separate administrative connector credentials from ordinary analyst accounts. A contractor who adjusts colors should not automatically inherit permission to manage all client data sources.
Client marketing and analytics team
Give designated client analysts editor access only when they need it and when the contract assigns them change responsibility. If they edit governed sources, establish a change-review process. Otherwise, provide viewer access plus a request channel for metric changes, new dimensions and data-quality issues.
Client executives and external stakeholders
Use viewer access and narrowly scoped distribution. Review download, print, copy, export and scheduled-email behavior according to the sensitivity of the data. Disabling a UI action is a useful control but not a complete data-loss prevention strategy. Authorized viewers may still be able to record information they can see.
Former staff, expired contractors and outgoing agencies
Revoke report access, source access, connector authorizations and scheduled delivery privileges as separate offboarding tasks. Removing an email address from a report is not equivalent to revoking an OAuth token or rotating a service credential. Review external groups, inherited permissions and copies as part of the exit checklist.
7. Decide report ownership before launch, not during a dispute
Agencies should agree at kickoff who is the enduring owner of each production asset. If the client pays for an ongoing business-critical dashboard, client-controlled ownership or an explicitly contracted organizational ownership model usually makes handover easier. The agency can retain editor access for maintenance. Exceptions, such as agency-owned reusable intellectual property, should be described in the agreement and should not be confused with ownership of the client’s underlying data.
Google documents ownership-transfer paths for eligible no-cost assets and organizational ownership with Pro team workspaces. The available transfer mechanism and restrictions depend on account type, organization and asset. Do not assume a cross-domain transfer will work simply because two accounts can share a report. Confirm the actual transfer path early using a disposable test asset and the identities that will participate in handover.
An ownership transfer also does not automatically rotate data credentials. Google notes that source credentials can remain intact after transferring a data source. A safe handover therefore includes reconnecting under an approved client identity where appropriate, testing all reports, revoking obsolete agency credentials and recording who will handle future connector failures. Refer to Google’s ownership-transfer guidance when designing the process.

8. Design credentials for continuity and least privilege
A production report that relies on a single employee’s personal Google account or third-party advertising login is an avoidable continuity risk. Prefer client-approved organizational identities and connector authorizations with clear ownership, documented renewal steps and least-privilege access. Where a connector supports only interactive user OAuth, document the account holder, recovery route, consent scope, token expiry or revocation triggers and fallback procedure.
Never share passwords by email to solve ownership. Use the identity provider’s supported delegation, transfer or administrator processes. Service-account credentials may be useful for supported warehouse connections, but they do not make a third-party marketing connector magically compatible. Maintain a credential inventory separate from the report inventory, with owners, renewal dates and an incident escalation contact.
When evaluating specialized integrations, Actiknow’s Looker Studio connector offerings illustrate the difference between connecting a platform and governing the resulting reporting estate. Connector availability is an integration capability; permissions, client isolation, credential ownership and report lifecycle still need an operating design.
9. Free versus Pro: choose for the operating model
The no-cost edition can be entirely appropriate for a small agency or a handful of low-risk client reports when owners are named, access reviews are performed and handovers are tested. Pro becomes more relevant when organizational ownership, team workspaces, workspace-level access management and stronger continuity are recurring requirements. An upgrade is not a substitute for data isolation, quality checks or connector monitoring.
Avoid making the decision solely on report count. Ten highly sensitive client reports with frequent team turnover may require more governance than 200 internal exploratory charts. Compare the total cost of administration, handover risk, licensing, connector support and incident response. If Pro is under consideration, evaluate a small set of representative clients and test external sharing, client ownership and team workspace behavior against the actual contract.
For a complementary explanation of reporting design and operational controls, Actiknow’s business intelligence services page describes dashboard development, data integration and ongoing reporting support. Those services can support the delivery process, but governance responsibilities must still be agreed between agency and client.
10. Build a control checklist for every production dashboard
Before launch, require evidence for the following checks: correct client source connections; approved KPI definitions; no cross-client fields or sample records; correct credential mode; named business and technical owners; minimum necessary viewer and editor lists; approved external sharing; expected refresh behavior; tested export and copy settings; reconciled totals; scheduled delivery recipients; incident contact; and documented offboarding.
A simple release record can contain the report URL, template version, source IDs, deployment date, reviewer, reconciliation result, access-test result and any accepted exceptions. Do not rely on a screenshot of the dashboard as evidence that the underlying permissions are safe. A reviewer should inspect the actual data-source and sharing configuration.
For a higher-risk report, test at least three identities: a permitted client viewer, a permitted client editor and an unauthorized user or user from another client. Record what each can open, query, copy, export and receive by email. Repeat these tests after major changes to credentials, data-source design, sharing policy or tenancy architecture.
11. Establish a maintenance and incident model
A dashboard can be perfectly permissioned on launch day and become unsafe or unreliable six months later. New client staff join, old staff leave, advertising tokens expire, source schemas change, campaigns are renamed and marketing definitions evolve. Assign a service owner for each dependency and review critical reports on a recurring cadence.
Track operational signals that matter to clients: freshness against a stated expectation, failed refreshes or connector errors, source-to-dashboard reconciliation, unexpected changes in audience, credential expiry, report performance and unreviewed template changes. A green dashboard load indicator does not prove that the numbers are current or that the report is appropriately shared.
For incidents, distinguish data exposure from reporting interruption. If a client can see another client’s data, immediately restrict access to the affected asset, preserve evidence, identify the source of the exposure and follow contractual and legal notification procedures. If a connector simply fails, isolate the failed dependency, communicate data freshness and restore through an approved credential or integration path. These are different severities and should not share the same generic ticket workflow.
12. A four-week rollout for agencies
Week 1: inventory and risk classification
Identify all client-facing production reports, data sources, connector accounts, owners and recipients. Rank them by sensitivity, business criticality, number of editors, external distribution and dependency on personal accounts. Select two representative clients for a pilot: one simple account and one with several data sources or stakeholder groups. Do not attempt to redesign the entire estate before understanding where the risks actually are.
Week 2: establish the client boundary
Choose client-specific data sources or enforced warehouse access, assign owners and document credential modes. Build an access matrix and separate agency development rights from client viewer rights. Validate that an unauthorized identity cannot retrieve another client’s data, including through copied reports, edits, exports and scheduled delivery.
Week 3: standardize templates and handovers
Create a clean master template, a deployment checklist, a KPI dictionary and a change log. Move pilot reports to the agreed ownership model. Test report copies, source reconnection, permission changes, connector renewal and the full handover sequence. Document any limitations that cannot be fixed immediately and the compensating controls.
Week 4: release, train and measure
Reconcile production metrics with client source systems, review access with client stakeholders and train agency staff on the release process. Define monthly or quarterly access reviews according to risk. Track whether new client deployments use the approved template and whether any production report lacks a named owner or tested exit path.
13. What agency leadership should measure
Governance is working when the reporting operation becomes less dependent on individual employees and less likely to expose client data. Useful measures include the percentage of production reports with named owners, percentage with documented credential modes, percentage passing cross-client access tests, number of orphaned assets, number of overdue access reviews, mean time to repair failed connections and percentage of handovers completed without business interruption.
Do not optimize only for the number of dashboards produced per analyst. Fast delivery that requires emergency reconstruction during client offboarding is not efficient. A more meaningful unit of performance is a validated, supportable client reporting deployment with clear accountability.
14. Executive decision framework
Use a client-specific report and data-source pattern when confidentiality is important, source systems are distinct and the agency has manageable client volume. Use client-specific reusable sources when multiple reports share governed metrics. Consider a warehouse-backed, multi-tenant model only with enforceable upstream permissions and negative security tests. Consider Pro when organizational ownership and team workspace administration solve an observed continuity or collaboration problem. Keep the visual template separate from the data authorization in every pattern.
The design is successful only if it survives three ordinary events: a new client is onboarded without access to an existing client’s data, an analyst leaves without breaking refreshes, and the client changes agencies without losing access to its own reports or sources. If any of those events requires an improvised workaround, the governance model is incomplete.
Frequently asked questions
1. Should the client or agency own the Looker Studio report?
For business-critical client reporting, agree on a durable client-controlled or explicitly contracted organizational owner before launch. The agency can remain an editor. Separate ownership of the report from ownership of the source account, credentials and reusable template intellectual property.
2. Does giving someone viewer access to a report give them access to the underlying data?
Not necessarily. The data source’s credential mode determines whether the report queries with the source owner’s authority or the viewer’s own authority. With owner’s credentials, a viewer may see report data without direct access to the original source. Review both asset permissions and query credentials.
3. Can we use one data source for all clients and filter by client name?
Do not treat a report filter as the only isolation control. Use separate sources or enforce tenant restrictions upstream, then test with users who can edit, copy and export. Shared multi-tenant sources require a stronger security design than a client filter in a chart.
4. Are reusable data sources always better than embedded ones?
No. Embedded sources simplify contained reports, while reusable sources help centralize definitions and connection management. Choose based on the editor model, client boundary, template deployment and change-control needs. Both require credential and permission review.
5. Does copying a report copy its reusable data sources?
No. Google documents that reusable data sources are not automatically copied or shared with the report. The deployment procedure must connect the copy to the intended client’s approved source and verify the result.
6. Can Looker Studio ownership be transferred between agency and client domains?
Transfer rules depend on the account type, organization and asset. Google documents restrictions for Workspace and Cloud Identity organizations. Test the actual transfer path before agreeing to an exit clause that depends on it.
7. Will changing report ownership also change the connector credentials?
Not automatically. Credentials can remain associated with the previous authorized identity. After transfer, verify the data source, reconnect under an approved identity where needed, and revoke obsolete authorizations only after the replacement is tested.
8. Is Looker Studio Pro necessary for every agency?
No. Smaller, lower-risk estates can operate with documented personal ownership and disciplined handovers. Pro deserves evaluation when shared organizational ownership, team workspaces and centralized administration materially reduce operating risk.
9. Should client stakeholders receive editor access?
Only when they have an approved need to change report structure or data-source behavior. Most executives need viewer access. For clients that require editing, define change approval, source ownership and responsibility for production incidents.
10. Are service accounts the best credential model for every connector?
No. Availability and configuration vary by connector and product. Use service accounts where supported and appropriate, but plan for user OAuth and third-party authorization requirements in marketing platforms. Document renewals and revocations either way.
11. How often should agencies review access?
Set a risk-based cadence, such as quarterly for critical external reporting and immediately after staff departures, contract changes or incidents. The exact interval is an internal policy decision. Review report roles, source permissions, credential owners, delivery recipients and copies.
12. What should a client receive when the agency relationship ends?
At minimum: the agreed report assets, ownership or access transfer, data-source inventory, credential transition plan, KPI definitions, refresh documentation, authorized-user list, support runbook and confirmation that the outgoing agency’s unnecessary access has been revoked. Validate continuity before closing the engagement.
A practical next step
Start with your ten most business-critical client dashboards. Identify the owner, the credential holder and the security boundary for each. Then test one client onboarding and one agency offboarding end to end. The gaps exposed by those exercises will tell you whether you need better procedures, a different data architecture or additional platform administration.
Actiknow works on BI implementation, data integration and dashboard development. If you need help reviewing a client reporting estate or designing a controlled Looker Studio rollout, contact Actiknow to discuss the specific sources, access model and operational responsibilities involved.
Selected technical references
Google documentation: Data credentials; About data sources; Embedded data sources; Transfer ownership; Roles and permissions. Product features and administrative limits may change. Confirm the settings available in your edition and account before deployment.

