Governance of Identity

Everything else in this pillar is about configuring access correctly. Governance is about keeping it correct as thousands of people and workloads join, change roles, and leave over years. Without it, access only accumulates, nobody’s permissions ever shrink, and the environment drifts toward a state where everyone can reach everything and nobody can say why.

This is the operational, unglamorous half of identity, and it’s where most real-world identity risk actually lives. It’s also the placeholder the site structure marks for expansion.

Joiner / mover / leaver

The lifecycle every identity follows, and each transition is a distinct failure mode:

• Joiner: new identity, appropriate access. The failure is over-provisioning at the start: cloning a similar user’s access, which copies their accumulated cruft on day one.
• Mover: role change. The most-neglected transition and the source of the worst drift. People change teams and gain new access without losing the old, “privilege creep.” After a few moves, a long-tenured employee holds access matching no current role, and that over-entitled account is exactly what an attacker wants.
• Leaver: departure, all access revoked. Conceptually simple, operationally leaky: the forgotten SaaS app with local accounts, the personal access token, the shared credential, the access in a system that SCIM doesn’t reach.

Automation is the only thing that makes this reliable at scale, driven from an authoritative source (usually HR) through SCIM to downstream systems. The mover case is the one worth engineering hardest, because it’s the one manual processes handle worst: nobody files a ticket to remove the access they no longer need.

Entitlement review at scale

Access recertification, periodically confirming people should still have what they have, is a real control and a real failure mode. Done badly it’s rubber-stamping: a manager handed a list of 400 entitlements they don’t understand approves all of them in one click, and the review has produced a compliance artifact and zero security.

What makes review meaningful rather than theatrical:

• Risk-based scope: review privileged and sensitive access often and carefully; low-risk access rarely. Reviewing everything equally guarantees everything is reviewed carelessly, the same prioritization logic as everywhere else.
• Comprehensible context: “Sarah has prod-db-admin, last used 6 months ago” enables a decision; a raw role name does not.
• Highlight the anomalies. Surface the outliers: access nobody else on the team has, access never used, access that predates the person’s current role. Those are where the review earns its cost.
• Revocation that works: a review that flags access for removal and doesn’t reliably remove it is worse than none, because it manufactures false assurance.

Identity Threat Detection and Response (ITDR)

Because identity is the primary attack surface, it needs its own detection discipline, and it sits at the seam between this pillar and detection engineering. ITDR watches the identity fabric itself:

• Authentication anomalies: impossible travel, new-infrastructure logins, MFA fatigue patterns, token replay.
• Privilege changes: role grants, new admin assignments, trust-policy edits, especially outside a change process.
• Directory attacks: the Entra and AD techniques, enumeration, consent-grant abuse, sync-account misuse.
• Dormant-account activity: an unused account waking up is one of the higher-signal events in security, precisely because it should never happen.

ITDR overlaps threat hunting heavily: identity is the richest hunting ground, for the same reason it’s the richest detection surface. Most intrusions transit it, and the behavioral signals (a service account behaving interactively, a human account behaving programmatically) are exactly what rules struggle with and analysts catch.

Governance is where identity meets compliance

Joiner/mover/leaver, access review, and least privilege are control requirements in every compliance framework: SOC 2, ISO 27001, FedRAMP all mandate them. That’s an opportunity, not just an obligation: the identity governance you build for security produces the evidence an auditor needs, generated as a byproduct rather than assembled before each audit. Identity is where a lot of “compliance as code” actually pays off, because the access data is already structured and already being collected.

How it looks in practice

The working version is an IGA platform (Entra ID Governance, SailPoint, Saviynt) wired to the HR system as the single source of truth, with SCIM carrying the changes downstream. The tell of a good deployment is the mover wiring: a team change triggers an access reset against the new role’s baseline, not merely new grants on top of old ones, which is the only mechanism that actually reverses privilege creep. Recertification campaigns are scoped by risk, show last-used data next to every entitlement, and their revocations are verified closed rather than assumed. Dormant accounts disable themselves on a timer. And the program reports one number above all others, time from termination to full deprovisioning, because it is the metric that compresses joiner/mover/leaver, coverage and automation into a single honest figure.

Where this connects

Governance operationalizes AWS and Azure identity: PIM, Access Reviews and least-privilege generation are governance tools. It rests on SCIM provisioning, feeds detection through ITDR, and produces GRC evidence as a byproduct. And governing the lifecycle of autonomous agent identities — at machine speed and scale — is the open problem in Frameworks.