C1.ai to join Workday. Read the announcement →

Identity Governance Framework: How to Build One

By The C1 Team
Identity Governance Framework: How to Build One

An identity governance framework defines who or what should have access across an enterprise, who can authorize it, and how teams grant, review and remove it. It connects business policy to the permissions people, services and AI agents actually hold.

After an acquisition, the CISO asks whether a former subsidiary employee can still export customer data. HR knows when the person left. IT disabled their corporate login. The subsidiary's application owner knows whether the person also had a local account. Each team did its part, but nobody agreed on who would bring those facts together and act on the answer.

That agreement is the work of building an IGA framework. The enterprise needs common rules for consequential access decisions and a way for each application team to carry them out. A useful starting point is to ask what must be consistent across the company and what only the business or application owner can decide. That distinction shapes everything from the first access request to the report that goes to the CISO. Identity governance and administration

What belongs in an identity governance framework?#

Five questions bring the main components into focus. The answers should work across finance systems, cloud infrastructure and business applications, even when each system grants access differently.

ComponentQuestion the program must answerWhat teams can use
Access policyOn what grounds can someone receive or retain a permission?Eligibility rules, incompatible duties and exception terms
Roles and ownershipWho understands the permission, approves its use and carries out the decision?Named business and application owners, access roles and escalation paths
Identity and access lifecycleWhat starts, changes or ends an entitlement for each identity population?Trusted sources, change events and fulfillment instructions
Access reviewsWho can assess continued need and what happens when they revoke access?Review scope, reviewer context and remediation records
Reporting and oversightWhere is the policy failing, and who can fix it?Coverage, verified outcomes, overdue changes and exception owners

These components depend on one another. A transfer can trigger a change only if someone has defined which access belongs to the old role. A reviewer can make a sound decision only if they know what a permission allows and why the person received it. A report can distinguish a successful removal from an open risk only if someone checks what changed in the application.

Illustrative view of an identity governance program. The organization decides who owns each decision and how changes are carried out.

A request, a lifecycle event or a review can trigger an access decision. Identity and permission records help the owner assess it. C1.ai editorial illustration.

Decide what the enterprise owns and what stays local#

Start with a small set of decisions that every business unit will honor. Who can accept the risk of an exception? Which types of access need independent approval? Who can require an application owner to resolve a failed removal? Put names or accountable functions beside those decisions. “IT owns access” gives a program lead little to go on when Finance and Engineering disagree about an exception.

The program owner maintains common policy, assigns responsibility for areas without an owner and brings unresolved issues to leadership. Business owners define which duties justify access and which combinations create unacceptable risk. Application owners translate those rules into privileges and explain the effects of changing them. A manager can confirm an employee's responsibilities. An administrator can implement a change. Neither responsibility automatically includes authority to waive policy. IDPro's guidance on business and IT responsibilities

An access role is a different use of the word role. A support engineer role might bundle case lookup and account diagnostics in several applications. The support leader owns the business reason for that bundle. Each application owner can confirm what its permissions allow. When one application adds bulk export to an existing permission, somebody needs authority to revisit the bundle. Otherwise the business can keep approving the same familiar role name as the access behind it expands.

For a new business unit, a short onboarding record makes the division of responsibility concrete. Capture the business owner, the technical owner, the sensitive actions the application permits, the sources that establish eligibility and the means of observing effective access. Record who will do the work if automation is unavailable. A subsidiary can keep a different ticketing system and still meet the same enterprise requirement to verify removal.

Connect policy to the identity lifecycle#

People, contractors and workloads have different beginnings and endings. HR can establish when an employee joins or transfers. A contractor's sponsor or contract system may be the best source for an engagement end date. An engineering team can identify the workload a service account belongs to and the event that retires it. For each population, establish who can vouch for its status and what happens if that information is missing or stale.

Then determine what a change in status means for access. Basic access may follow from an approved job role. Sensitive privileges may require an application owner's approval, a check for conflicting duties or a time limit. A transfer can entitle an employee to a new role while leaving the old assignment in place through a local group. Checking only what the employee gained would miss half of the change.

Consider a single treasury employee moving to procurement. Finance says her new job still requires read access to invoices but no longer permits her to release payments. The payment-system owner finds that she holds the release privilege directly and through a treasury group. An administrator removes the direct grant and supplies a narrower read role before removing the group membership. The owner checks both results in the payment application. The same business rule can apply at another subsidiary even if its banking portal has entirely different permissions.

Exceptions need a lifecycle of their own. If Finance needs the employee to finish a handover, the person with exception authority can specify which transactions she may release, who will check them and when the arrangement expires. The end date gives the application team an action to take. A record of the original approval alone says nothing about whether the handover is still underway.

Design reviews around decisions people can make#

A review is a chance to catch access that ordinary lifecycle events missed. Give reviewers a business question they can answer. A manager may know whether someone still supports a customer but have little idea what the group labeled “tier 3 operations” permits. The application owner can explain the group's capabilities. For sensitive access, those two perspectives can be combined before someone decides whether to keep it.

Choose review scope and timing according to the exposure and the changes the business can observe. An expiring engagement can trigger a decision about one contractor. A periodic review can catch a department that routinely extends contractor access beyond the original project. Privileged production roles may warrant a narrower review with more context than everyday collaboration tools. Each choice should have an owner who can explain why the review is worth the time it asks of people.

Plan how a revoke decision reaches the system. For an integrated application, the team may submit a removal automatically and reconcile permissions afterward. A legacy application may require an operator to make the change and return a fresh permissions export. In either case, preserve what the reviewer decided, who acted and whether effective access changed. These distinctions matter when a completed ticket and the application tell different stories.

Give leadership a view of what works#

A completion rate answers whether reviewers finished their assignments. It tells leadership little about systems outside the campaign or a revoked privilege that an employee still holds. Report coverage alongside results. A useful coverage measure is how many in-scope applications have an accountable owner and a current permissions inventory, with the total number in scope shown beside it. Document which systems are missing and who can bring them into the program.

For removals, compare the changes verified by their deadlines with all removals due in the period. If 18 of 20 were verified on time, the result is 90 percent. Keep the two late cases visible with their age and owner. One might be waiting for a reviewer with authority to decide. The other might have failed because a connector cannot remove an inherited grant. Only the second calls for a change to fulfillment.

Use these results to make decisions. Repeated exceptions for one support role may mean that the role is poorly designed. A backlog concentrated in the same legacy application may justify integration work. A business unit missing from the inventory needs a sponsor and a plan for coverage. Taken together, those observations tell a program lead where the next investment would improve access governance.

Include service accounts and AI agents in the same program#

A non-human identity needs a purpose, an accountable owner and a way to establish what it can actually do. A service account may outlast the engineer who created it. A short-lived agent can run under authorization that survives many individual sessions. Put responsibility for that authorization in the program while the workload is active, and arrange a reassignment decision when its sponsor leaves. Agentic identity governance

For example, an invoice reconciliation agent may need to read invoice and payment status. The Finance owner can define that business purpose, and the technical owner can identify the identity, grants and tools that make it possible. If it acts for a user, capture whose authority it uses and the scope of that delegation. Adding a tool that can initiate a payment gives the owners a concrete reason to reassess the agent's authority.

Governance also needs a clear boundary with runtime enforcement. A review can determine whether the agent should retain a permission. The application or gateway must enforce which actions it can take during a session. At retirement, withdraw persistent grants and delegations, check whether the agent can obtain a new credential, and handle existing sessions according to the system's capabilities. Looking only at the expiry of the last token can leave the underlying authorization intact.

Build from the gaps you can demonstrate#

C1's Path to Identity Maturity gives an IAM lead a way to assess progress. Its six stages, numbered 0 through 5, cover blind spots, discovery, hygiene and rationalization, governance, just-in-time access and zero standing privilege, and autonomous identity. Apply its questions to a particular population or business area. Production engineers may already use temporary elevated access while a newly acquired unit still needs to discover its local accounts.

Choose an initial scope with both a business consequence and a way to observe the result. For a team struggling with visibility, that might mean identifying every route to administrative access in production and assigning an owner to each. For a team that already has good inventories, it might mean correcting a recurring failure after job transfers. Test the decision through a real change, then identify what the next application can reuse and which permissions or owners will differ.

Organizations that already report against NIST's Cybersecurity Framework can use Current and Target Organizational Profiles to record their progress against selected CSF Core outcomes. For example, PR.AA-05 addresses policy, enforcement and review of permissions, including least privilege and separation of duties. Evidence from an access review can help a team assess that outcome. The team still sets its own scope, delegates authority and chooses how to close each gap. NIST CSF 2.0

When the manual program is ready to scale#

Look for work that teams repeat even after they agree on the right decision. One analyst pulls group memberships, another reconciles them with HR, and an application owner receives a spreadsheet with the same questions as last quarter. As applications multiply, that routine can consume the time people need to investigate unusual access. Automating a well-understood decision lets the owner concentrate on the exceptions that still require judgment.

Treasure Data started with sporadic, mostly AWS-focused spreadsheet reviews. Its team established a quarterly process spanning roughly 15 systems before adopting C1. That manual work exposed exactly where the effort went, including collecting exports, reconciling user lists and chasing answers. With C1, the team could refine which grants went to reviewers instead of asking them to assess every permission the same way.

C1 Identity Governance connects lifecycle changes, policy-based access requests and reviews across supported systems. The business can define who receives ordinary access, which requests require approval and when elevated access should expire. In a C1 access review, the team can route a decision to the appropriate manager or application owner, move a revocation into a supported remediation workflow and track whether the action reached its destination. The team can keep a manual fulfillment path for an application that requires one.

When modernizing an existing identity program, use one of its difficult cases to decide what to connect next. If one team spends days chasing decisions, better routing may be its highest return. If every decision arrives on time but access survives through a group, invest in the target-system integration and the ability to check effective permissions. Both teams can use the same platform and still have good reasons to start in different places.

Frequently asked questions#

What is an identity governance framework?#

An identity governance framework defines how an organization decides who or what should have access, assigns responsibility for those decisions and verifies what people and workloads can actually do.

What are the main components of an IGA framework?#

An IGA framework covers access policy, roles and ownership, identity lifecycle changes, access reviews, and reporting. Together they connect business decisions to permissions in individual systems.

Who owns identity governance?#

An enterprise program owner coordinates governance. Business owners define who needs access, application owners map those needs to permissions, and designated operators or integrations carry out approved changes.

Where should an identity governance program start?#

Choose a consequential permission with a known business owner and a way to check the result. Use a real event, such as a transfer or contract expiry, to test the policy across the systems involved.

How does an IGA framework apply to AI agents?#

Assign an owner to the agent's purpose, record its permissions and delegated authority, and review changes to its tools or scope. When the agent retires, withdraw its grants and verify that access ends.

Evaluate C1 with a case from your own program#

Bring an application, its access policy and a recent change your team found difficult to verify. Ask to see which decisions your owners would make, which work C1 could automate and how you would know that the resulting access matched policy. Explore C1 Identity Governance