Salesforce user management often begins as a straightforward administrative process. A new employee joins, receives a profile, gains a few permission sets, and starts working. When responsibilities change, access is adjusted. When the employee leaves, the account is deactivated.
That model works until the organization becomes more complex.
As Salesforce expands across business units, geographies, acquired companies, external partners, applications, and integrations, access decisions begin accumulating faster than teams can evaluate them. Users receive additional permissions for temporary projects. Administrators create new permission sets to solve urgent requests. Regional teams interpret corporate policies differently. Integrations retain access long after their original purpose changes.
Eventually, the organization can answer whether a user has access, but not always why the user has it, whether it is still necessary, or what that access makes possible.
That is where user management stops being an administrative function and becomes an enterprise risk.
We’ll explore these seven aspects of Salesforce user management:
- Access Accumulates Faster Than It Is Removed
- Job Titles Become a Weak Proxy for Business Need
- Permission Sets Multiply Without a Coherent Model
- Joiner, Mover, and Leaver Processes Drift Apart
- Integrations and Non-Human Identities Expand the Problem
- Periodic Reviews Become Compliance Exercises
- Governance Must Connect Identity, Configuration, and Activity

1. Access Accumulates Faster Than It Is Removed
Most access problems are not caused by a single reckless decision. They are caused by hundreds of reasonable decisions made at different times.
An employee begins in sales operations, moves into revenue strategy, helps with a data migration, and later joins a regional transformation project. Each assignment adds access. Few transitions trigger a complete review of what the employee already has.
This creates permission accumulation. Users retain capabilities connected to previous roles, completed projects, temporary responsibilities, or legacy business processes. The individual permission assignments may appear harmless. Combined, they can provide far broader access than anyone intended.
Complex organizations are particularly vulnerable because responsibility for access is distributed. Human resources knows the employee’s title. The manager knows the current responsibilities. The Salesforce team understands the platform permissions. Security understands the risk. No single group necessarily sees the complete picture.
The result is an access model that reflects a user’s employment history rather than the work the user performs today.
2. Job Titles Become a Weak Proxy for Business Need

Many organizations design access around roles, such as sales representative, service manager, analyst, or administrator. That provides a useful starting point, but job titles rarely capture the full context required for precise authorization.
Salesforce recommends granting permissions according to the tasks users perform rather than relying only on job titles. It also recommends starting with minimum access and adding capabilities through permission sets and permission set groups.
The important shift is from title-based access to capability-based access. Instead of asking, “What should a sales manager receive?” organizations should ask, “What specific actions must this person perform, on which data, under which conditions?”
3. Permission Sets Multiply Without a Coherent Model
Permission sets are designed to make access more modular. In practice, that flexibility can produce a new form of complexity.
Teams create permission sets for departments, applications, projects, exceptions, integrations, releases, and one-time requests. Naming conventions vary. Similar sets are duplicated. Some contain a single permission while others contain broad collections of capabilities. Administrators may hesitate to remove an assignment because its purpose is unclear.
Salesforce itself identifies ad hoc, redundant, heavily duplicated permission sets as an authorization anti-pattern. Its Well-Architected guidance recommends aligning permission sets and permission set groups with defined business capabilities and documenting the relationship between users, personas, and permitted access.
Modularity alone does not create control. Permission sets need architecture.

4. Joiner, Mover, and Leaver Processes Drift Apart
Organizations often focus heavily on provisioning new users. New hires need access quickly, and delays are visible. Deprovisioning and role changes receive less consistent attention.
The “mover” process is usually the weakest. A user changes departments, regions, projects, or responsibilities but remains active in Salesforce. New access is added to support the new position, while old access remains because no one has clearly defined what should be removed.
Leaver processes can also become fragmented when users exist across multiple Salesforce orgs, sandboxes, identity providers, connected applications, and partner environments. Deactivating one account does not necessarily terminate every path into Salesforce data.
Salesforce supports user access policies that can automatically grant or revoke permission sets, permission set groups, licenses, queues, and group memberships when user attributes change. Automation can improve consistency, but only after the organization establishes reliable rules. Automating a poorly defined access model simply applies the wrong decisions faster.
Lifecycle controls should treat every material role change as an access event, not just an HR update.
5. Integrations and Non-Human Identities Expand the Problem
Salesforce access is no longer limited to employees signing in through a browser. Integrations, APIs, automated processes, bots, data tools, and AI agents all require identities and permissions.
These accounts are frequently over-permissioned because broad access is easier to configure and less likely to interrupt a business process. They may also be shared across multiple integrations, making it difficult to determine which system performed a specific action.
Organizations should create a unique integration user for each integration and limit that identity to the minimum access required. This improves both security and traceability. When several systems share one highly privileged account, teams lose the ability to isolate activity, revoke access selectively, or determine the blast radius of a compromised credential.
Every non-human identity should therefore have a defined owner, documented purpose, approved permissions, credential management process, and retirement trigger.
6. Periodic Reviews Become Compliance Exercises
Many organizations conduct quarterly, semiannual, or annual access certifications. Managers receive a list of users and permissions and are asked to approve or reject them.
The process creates evidence that a review occurred. It does not always demonstrate that meaningful decisions were made.
Long permission lists are difficult to interpret. Technical labels provide little business context. Reviewers may not know what a permission enables or whether it is inherited through another assignment. When nearly everything was approved during the previous review, approving it again becomes the path of least resistance.
Effective reviews should focus attention on risk, not simply present every assignment with equal weight. Elevated administrative privileges, broad data access, dormant accounts, unusual combinations of permissions, external users, and high-impact integration identities should receive deeper scrutiny.
Reviewers also need context: why the access was granted, when it was last used, who owns the underlying capability, and how the assignment compares with peers performing similar work.

7. Governance Must Connect Identity, Configuration, and Activity
User management is often treated as a provisioning problem. In reality, it spans three connected questions:
- Identity: Who or what is accessing Salesforce?
- Authorization: What can that identity see and do?
- Behavior: Is the identity using that access as expected?
An account can be legitimate while its permissions are excessive. A permission can be approved although the resulting activity is unusual. A well-designed access model can still deteriorate after configuration changes, business restructuring, or acquisitions.
Organizations need a repeatable way to compare intended access with effective access and observed behavior. Tools such as AutoRABIT Guard can support this effort by helping teams assess Salesforce configurations, permissions, access pathways, and security posture across changing environments. This gives security and platform teams greater visibility into where access has expanded, where configurations have drifted, and where controls may no longer reflect policy.
Technology alone does not solve the governance problem. Effective oversight still requires cooperation across security, identity, Salesforce administration, compliance, human resources, application ownership, and business leadership.
User Management Is an Architecture Decision
Salesforce user management rarely breaks down because organizations do not care about security. It breaks down because their access models fail to evolve at the same pace as the business.
Organizations need to define access through business capabilities, apply least privilege by default, automate lifecycle changes where the rules are reliable, separate integration identities, and continuously evaluate whether effective access still matches legitimate need.
The central test is simple: Can the organization explain who has access, what that access enables, why it is required, and when it should end?
When the answer requires days of investigation, user management has already become a strategic risk.