An entitlement review consultant is the engagement you buy when the review is due and the grants still live in five systems with five export shapes. The work product is a single grant table, a named reviewer for each row, a recorded decision, and a file an examiner can sample. Rutagon runs that engagement from Alaska against the accounts and tenants you already operate. We do not become the auditor, and we do not mark a row approved because the due date arrived.
A full-time hire owns the program after the quarter. The engagement owns the systems in scope and the window in the statement of work. Many clients keep the collectors when the window closes. The review itself still needs a human decider for each grant. That decider is someone in your company who knows why the access exists. Our name on the decision is a defect.
What an entitlement review consultant hands you
The first deliverable is the source list. Each source defines what an entitlement is in that system. In IAM Identity Center it is an account assignment and a permission set. In a Git organization it is a team membership and a repository role. In a data warehouse it is a role grant. In an internal admin tool it is a column the product team named when the tool was first built. If the source cannot say what a grant is, it is not ready to review, and it stays on the open list instead of being skipped quietly.
The second deliverable is the normalized table. Every row has the same fields, so a reviewer does not learn a new spreadsheet per system.
source, account, subject, entitlement, collected_at
reviewer, decision, reason, decided_at
remediation, request_id, rechecked_at The third deliverable is the reviewer map. Managers review their people where the directory says who the manager is. System owners review shared accounts, service roles, and anything with no human subject. A row with no reviewer is not distributed. It is an exception with a due date. Spreading those rows across a channel and calling them assigned is how the campaign reaches the due date with a third of the rows still blank.
The fourth deliverable is the decision log and the remediation list. Approve, revoke, and acknowledge-with-a-reason are the decisions we accept. A blank cell is pending. Pending is not approve. Revoke becomes a remediation item with the API or the admin action that removes the grant, a request id, and a recheck that reads the grant again and shows it gone. A revoke that exists only as a cell in the sheet will still be there when the examiner samples it. The neighboring engagement shape, when the gap is the automation program rather than this window's packet, is access governance consulting. When the need is a person to implement the campaign tooling, the closer description is access review implementation.
Scope, systems, and the rows people hide
Scope is a list of systems and a window, written down before collection starts. Adding a system in the last week produces an empty review for that system or a rushed one. We would rather cut a system from scope, with the cut visible, than invent completeness. Out-of-scope systems are named. An unnamed system that an employee uses for production access will be the one the examiner finds.
Collection is read-only. We use the API where it exists and an export where it does not. A screenshot of a settings page is accepted for a system that has no export, and it is labeled as a screenshot, with the date and the person who captured it. It does not get reformatted into a row that looks like an API pull. The internal tool with no SCIM endpoint is usually the one that matters, because the directory review already happened and this grant did not. Custom applications are in scope when they can change production or customer data. A marketing tool with no production path can be listed and deferred. The deferral is a line in the pack, not a missing page.
Shared logins, break-glass roles, and contractor accounts each have a rule before the campaign opens. Shared logins need a human owner and a member list. Break-glass needs an owner and a reason it was unused, or a revoke. Contractors need an end date. A row that matches none of the rules stays unassigned until you pick a rule. We do not default contractors to "approve" because their sponsor is busy.
Privileged entitlements are reviewed by a named owner of that privilege, not only by the subject's manager. A manager can confirm the person still works here. The manager often cannot confirm that the person still needs the production admin path. Those two questions are two rows, or one row with two deciders. Collapsing them produces a review that looks finished and does not answer the privilege question.
Evidence, remediation, and what we will not sign
The evidence pack has the source list, the collected table, the reviewer map, the decisions, the exception list, and the remediation rechecks. It is the pack we drop into the GRC platform you already run, as an attachment or through the API that platform already exposes. We do not ask reviewers to click the same decision a second time in the GRC if the decision log is complete. Two campaigns create two answers for one grant.
Remediation can be yours or ours. When it is ours, we execute only the revokes a named reviewer recorded. The recheck is part of the item. A ticket that says "removed" with no second read of the grant is open. When remediation is yours, we still want the recheck result in the pack before we call the window closed. A decision log full of revokes and a directory that still has the grants is the failure mode examiners already know.
We will not approve rows on your behalf. We will not backdate a decision to fall inside the window. We will not drop a source because its export was inconvenient. We will tell you the row count per source before reviewers start, so a collector that returned ten grants for a system you know has hundreds is fixed before people spend a day on the wrong extract.
Rutagon's reviewers for this work are the client's reviewers. Our engineers build the collectors and the pack. On systems we operate ourselves, including a production SaaS platform, entitlements are collected the same way: one row shape, a named decider, a recheck after revoke. That is the standard we will hold the engagement to. The engagement is security automation applied to one review window, with the humans left in the decisions. A platform license that already sends reminder email does not replace the source list. If the reminder tool can see the grant, we collect from it. If it cannot, we collect beside it.
A useful first conversation is the source list and the date the window closes. Systems you cannot name yet are part of that conversation, not a reason to wait until the list feels complete. The list gets less complete in the last week, not more.
Frequently Asked Questions
Who decides to approve or revoke a grant?
A named person in your organization. Managers decide for their people where the question is whether the person should have access at all. System owners decide for shared accounts, service roles, and privileged paths. Rutagon records the decision and can execute a revoke you already made. A row we mark ourselves is a row we reopen.
What if a system has no API and no SCIM endpoint?
It stays in scope when it can affect production or customer data. Collection is an export or a labeled screenshot with a date and a capturer. The row is marked so it cannot be mistaken for an API pull. Dropping the system because the connector catalog does not list it is the outcome this engagement exists to avoid.
How is this different from hiring a full-time access review owner?
A hire owns the program going forward: collectors, campaign calendar, and the next window. The engagement owns the systems and the window in the statement of work, and it leaves you the collectors and the pack. Teams use it to finish the review that is already late, then decide whether the program needs a seat.
What does the examiner actually receive?
The source list, the normalized grants, the reviewer for each row, the decision and timestamp, the exceptions, and the remediation rechecks. Totals and reminder screenshots are optional context. The sample is pulled from the decision log, and the recheck is what shows a revoke left the system.
Book a technical call. Contact Rutagon at rutagon.com/contact, 907-841-8407, or contact@rutagon.com.