By the end of this guide you’ll have a complete, repeatable process for:
Removing users safely from your AI ad production tools
Designing role-based access controls that match how creative teams work
Maintaining audit trails so you always know who changed what and when
Balancing governance with fast, high-volume ad production
We’ll work through each piece as a practical workflow you can implement in any AI ad platform, using Viewst’s approach as a concrete example.
Prerequisites
Before you design or refine your user deletion and access controls, make sure you have:
A defined org structure: teams, brands, clients, and workspaces mapped out.
Documented roles: e.g., Creative Director, Designer, Producer, Marketing Manager.
A basic data retention policy: how long you keep campaigns, assets, and logs.
Identity and SSO strategy: at minimum, who owns provisioning and offboarding.
Legal and compliance input: especially if you operate in GDPR/CCPA regions.
With those in place, you can design controls that are both realistic and enforceable.
1. Separate deprovisioning, data retention, and deletion
The most common failure in creative tools is treating “delete user” as one button that fixes everything.
Regulators like the UK ICO make two key points about deletion:
If you can delete, you should delete in line with your retention schedule.
If system constraints prevent deletion, you must restrict access and move records out of normal use.
Actionable steps
Define three distinct workflows:
Account deprovisioning – remove access and log the event.
Asset retention – keep or archive creative assets according to policy.
Legal holds – mark data that cannot be deleted yet.
Create a simple decision tree:
Is this a normal offboard? → Deprovision user, retain assets per policy.
Is there a “right to be forgotten” request? → Delete personal data where possible, pseudonymize elsewhere.
Is there a legal/compliance hold? → Deprovision user, flag assets as locked, don’t delete.
Implement these as admin actions in your platform:
Deactivate userTransfer ownership(campaigns, folders, brandbooks)Delete personal profile data(where legally required)
Example in an AI ad platform like Viewst
Deactivation: removes login access, keeps their work connected to campaigns.
Ownership transfer: reassigns campaigns and brand systems to a new owner.
Data minimization: replaces personal identifiers in logs with pseudonymous IDs when retention periods expire.
Common failure
Relying on a single “delete user” action that either:
Leaves orphaned campaigns with no clear owner, or
Keeps the user technically active because deleting them would break the system.
The fix is to separate access from data and design workflows for each.
2. Design role-based access control (RBAC) around real creative workflows
NIST identifies RBAC as the predominant model for advanced access control because it reduces admin overhead and complexity.
For creative teams, RBAC must map to how work actually happens.
Actionable steps
List your core roles (keep it under 8 for manageability):
Creative Director
Design Lead
Designer / Motion Designer
Creative Producer / Project Manager
Marketing Manager / Campaign Manager
Localization Manager
External Agency or Freelancer
Define responsibilities per role:
What can they create? (campaigns, templates, brandbooks)
What can they edit? (layouts, animations, copy)
What can they approve? (final creatives, versions)
What can they export or publish? (HTML5, GIF, video, tags)
What can they administer? (users, permissions, brand systems)
Translate responsibilities into RBAC permissions:
Designer:
Can edit creatives and animations in assigned workspaces.
Cannot change brandbooks or global styles.
Cannot manage users.
Creative Director:
Can approve campaigns.
Can override layouts and animations.
Can manage brand systems in specific workspaces.
Admin:
Full user and workspace management.
Access to audit logs and security settings.
Apply least privilege:
Start with the minimum access a role needs to meet its responsibilities.
Add exceptions through controlled processes, not ad-hoc.
Viewst-style permissions structure example
In a platform like Viewst, you might implement:
Roles: Admin, Brand Owner, Creative Lead, Designer, Reviewer, Viewer.
Capabilities per role:
Admin: user management, workspace creation, log access.
Brand Owner: brandbooks, asset libraries, approval workflows.
Designer: canvas editing, AI tools, Smart Resize, export drafts.
Reviewer: comment, approve, reject, no editing.
Viewer: view-only access for stakeholders.
Common failure
Creating too many custom roles or per-user exceptions, which quickly become unmanageable and lead to misconfigurations.
Stick to a small, clearly documented role set and use workspaces/groups to handle visibility.
3. Use workspaces and groups to control what people can see
RBAC alone controls what users can do, but high-volume creative teams also need controls for what users can see.
Storyteq and other enterprise tools use a roles + groups pattern:
Roles = what
Groups/workspaces = where
Actionable steps
Define workspaces around clear boundaries:
By client (e.g., “Client A”, “Client B”).
By brand/line of business.
By region or market.
Map groups to real teams:
Client-specific agency squads.
Regional marketing teams.
Localization vendors.
Combine roles + groups for access control:
A Designer in the “EMEA” group sees only EMEA workspaces.
A Creative Director has read access across regions, edit access in their primary region.
Freelancers are isolated to one client workspace.
Enforce default scoping:
New campaigns inherit the workspace’s default visibility.
Guest access expires automatically or is reviewed quarterly.
Viewst-inspired pattern
Organizations (your company)
Workspaces (client or brand)
Projects/Campaigns (specific initiatives)
Users get:
A role at org or workspace level.
Membership in specific workspaces.
This keeps creative data compartmentalized while allowing shared workflows where needed.
Common failure
Granting org-wide access “just to keep things simple,” which increases breach impact and makes it impossible to isolate sensitive campaigns.
Use workspaces and groups as your default boundary.
4. Make onboarding and offboarding a first-class workflow
Internal errors cause more trouble than malicious insiders. The 2025 Verizon DBIR found that internal actor breaches had more miscellaneous errors than privilege misuse, at roughly a 2:1 ratio.
That means your onboarding and offboarding processes must be operationally simple.
Actionable steps
Standardize onboarding:
All users come through a central identity provider (IdP) like Okta.
Assign default roles and workspace memberships based on job function.
Require SSO for all logins.
Design a deprovisioning checklist:
Disable account in IdP.
Automatically revoke access to connected ad tools.
Transfer campaign ownership.
Confirm removal from high-risk groups (e.g., AI model admin, export admins).
Align with identity tools:
Use deprovisioning reports (like Okta’s) to ensure all connected apps, including creative platforms, are covered.
Make offboarding events visible in both the IdP and the ad platform’s audit logs.
Formalize approvals:
Require manager or creative ops approval for user creation and deletion.
Log every deprovision action with time, actor, and reason.
Viewst-style process
SSO integration: Admins manage users centrally; Viewst trusts the IdP for identity.
Deactivation workflow:
User deactivated in IdP → Viewst account auto-blocked.
Admin transfers projects to a new owner via a guided wizard.
Audit log records deactivation and ownership changes.
Common failure
Manual, ad-hoc offboarding where designers lose access to some tools but stay active in others for months.
Automate deprovisioning wherever possible and make it auditable.
5. Implement detailed audit trails and content logs
Audit trails are your safety net.
IBM’s 2025 research found that 63% of breached organizations lacked AI governance policies, and 97% of AI model or app breaches happened in organizations without proper AI access controls.
Without logs, you can’t prove governance even if your policies are solid.
Actionable steps
Define what you must log:
Administrative actions: user creation/deletion, role changes, SSO configuration, integrations.
Content actions: create/update/delete ads, animations, templates, brandbooks.
Access events: logins, failed logins, export/publish events.
Specify minimum log fields:
Date and time (UTC)
User identifier (and mapped role)
Action type (create/update/delete/export)
Object type (campaign, asset, brandbook, user)
Object ID or name
Old vs new values for critical fields (e.g., status, owner)
Make logs accessible:
In-app audit log UI for admins.
Export or streaming API for SIEM tools.
Protect logs from tampering:
Only security admins can modify log retention.
End-users cannot delete or alter logs.
Adobe-style benchmark
Adobe’s enterprise tools log actions such as create/update/move/delete and provide both content logs and admin audit logs, which is the standard to emulate.
Viewst-inspired implementation
Audit log for:
User lifecycle events (create, deactivate, role change).
Brand and workspace changes (permissions, ownership).
Export/publish actions (who shipped what, where, and when).
Content log for:
Creative edits: layout changes, animation tweaks, copy updates.
AI-assisted operations (e.g., Smart Resize, AI Designer runs).
Common failure
Only logging logins and ignoring content changes, which leaves you unable to answer “who changed this ad?” when there’s a compliance or brand incident.
6. Govern AI access, prompts, and exports explicitly
AI access control is now a specific risk category. IBM reports that 13% of organizations have experienced breaches involving AI models or applications.
In AI ad production tools, governance must include:
Who can run AI operations
Which models or presets they can use
How prompts and outputs are logged
Actionable steps
Define AI-specific permissions:
Access to AI Designer or layout generation.
Access to Smart Resize and Instant Animator.
Authority to create or edit AI templates/presets.
Control model and data access:
Limit who can connect external AI services or APIs.
Restrict integration keys to trusted admins.
Log AI interactions:
Record which AI tools were used on which assets.
Capture prompts and key parameters where privacy allows.
Link AI runs to specific users and campaigns.
Set export/publish controls:
Separate permissions for “edit” vs “export/publish to network.”
Require approvals for certain placements or ad types.
Viewst-style AI governance
AI tools as features tied to roles:
Designers: full access to AI Designer and Instant Animator within their workspace.
Reviewers: no AI generation, view-only.
Creative Leads: can approve AI-derived templates into the brand system.
Export controls:
Only roles with “Publisher” or “Campaign Manager” rights can generate final HTML5 bundles and click-tag-ready exports.
Common failure
Treating AI features as just another design tool with no separate controls, leading to untracked model use and misaligned governance.
7. Align user deletion with GDPR and other privacy rights
If you operate in regulated regions, your user deletion process must accommodate rights like GDPR’s “right to be forgotten.”
ICO guidance suggests:
Delete personal data where possible.
If deletion isn’t feasible, restrict access and move records out of normal use.
Actionable steps
Catalogue personal data in your platform:
User profile details (name, email, role).
Collaboration data (comments, approvals).
Activity logs.
Define what gets fully deleted vs pseudonymized:
Delete: profile records, contact details, stored preferences.
Pseudonymize: authorship references in creative histories, logs beyond your retention window.
Design a standard response for “delete my data”:
Verify identity.
Run a deletion workflow across connected systems.
Provide a summary of what was deleted, retained, or pseudonymized.
Align with retention schedules:
Set clear log and asset retention windows.
Automate deletion or anonymization where possible.
Viewst-style approach
Admin-only personal data deletion tools.
Retention-based log handling, where logs are retained for a defined period and then minimized.
Asset-level governance so campaigns can outlive accounts without exposing personal data.
Common failure
Promising full deletion but keeping identifiable references in logs and comments, which can undermine compliance if not documented and justified.
8. Test, monitor, and iterate your governance model
Designing the model is only half the job. You need to validate that it works under real creative pressure.
Actionable steps
Run dry-run offboarding tests:
Pick a test user and simulate a full offboard.
Confirm access removal, ownership transfer, and logs.
Review audit logs regularly:
Monthly checks for unusual role changes or export patterns.
Quarterly sampling of content logs for compliance.
Survey power users:
Ask designers and producers whether access controls slow them down.
Identify friction points and adjust roles/workspaces.
Reassess annually:
Align with new regulations and internal policies.
Review whether your role structure still matches the org.
If you want a broader view of how feature design and controls support transparency, see this related in-depth guide: ad transparency in creative production and how feature design and user controls support it.
FAQ: Troubleshooting user deletion and access controls
How should I handle deleting users without breaking existing campaigns?
Treat deletion as deprovisioning plus ownership transfer.
First, deactivate the user so they can’t log in.
Then, transfer ownership of their campaigns, brandbooks, and workspaces.
Finally, delete or anonymize personal data as required by your policy.
This keeps campaigns intact while removing the person from the system.
What’s the best way to structure permissions for agencies with many clients?
Use a combination of:
Workspaces per client, each with its own projects.
Roles per job function (e.g., Designer, Creative Director, Client Reviewer).
Groups that map to client teams, restricting who can see each workspace.
Designers only see the clients they work on; leadership roles can have cross-workspace visibility where justified.
How detailed do audit logs really need to be?
At minimum, your logs should answer:
Who made this change?
What did they change?
When did they change it?
What was it before and after?
Capture this for user management, permissions, exports, and significant creative edits. Adobe’s approach—separate admin audit logs and content logs—is a useful reference.
How do I prevent ex-contractors from accessing my AI ad tools after their projects end?
Integrate your creative platform with an IdP like Okta.
Use time-bound access for contractors and freelancers.
Make offboarding part of your project closeout checklist.
Confirm deprovisioning in both the IdP and platform audit logs.
This aligns with identity market norms where offboarding is a measurable security control.
How does Viewst help with governance without slowing production?
Viewst is designed as an AI ad production platform with:
Role-based permissions for admins, creative leads, designers, and reviewers.
Workspaces that isolate client and brand campaigns.
Native HTML5 output that remains fully editable.
Integrated approvals and logs that keep governance in the workflow, not bolted on afterward.
If you want to see how this works in practice, you can explore the platform at viewst.com.

Victoria is the CEO at Viewst. She is a serial entrepreneur and startup founder. She worked in Investment Banking for 9 years as international funds sales, trader, and portfolio manager. Then she decided to switch to her own startup. In 2017 Victoria founded Profit Button (a new kind of rich media banners), the project has grown to 8 countries on 3 continents in 2 years. In 2021 she founded Viewst startup. The company now has clients from 43 countries, including the USA, Canada, England, France, Brazil, Kenya, Indonesia, etc.
