Cloud teams that set out to buy CIEM, or cloud infrastructure entitlement management, usually start with a demo that shows a long list of risky permissions. The harder question is whether the tool will help the team decide what each permission is for, who owns it and what should happen next. This buyer guide sets out the capabilities to evaluate, the questions to ask in a demo and the warning signs to watch for, so a CIEM evaluation measures governance outcomes rather than dashboard volume. Citadel Identity360 brings CIEM capability and identity governance into one platform, and the checklist below reflects how we think a cloud team should test any option, including ours.
If you need the background first, our article on entitlement sprawl in cloud and SaaS explains how excess permissions build up, and our earlier post on Citadel Identity360 for cloud entitlements explains the limits of provisioning on its own. This guide assumes the problem is understood and focuses on choosing the right tool.
CIEM in plain language for a buying team
A CIEM tool discovers, analyses and governs cloud entitlements, so a team can see who or what holds access to cloud resources, what that access allows and whether it is still justified. It does not replace the authorization engine inside each cloud platform. Native controls still enforce access; CIEM sits above them to make that access understandable and manageable.
Three terms keep an evaluation honest:
- Identity: the person, service account, workload or application that requests access.
- Entitlement: the permission grant itself, delivered through a role, a policy or an assignment.
- Permission use: what the identity actually does with the access it holds.
A capable tool keeps all three in view. A principal can hold a grant it never exercises, and a role name rarely reveals the full effective access behind it. If a product shows only one of these layers, the buying team will be left to join the rest by hand.
Decide what your cloud team must answer before any demo
Buyers get better demos when they arrive with the questions their own estate raises. Before shortlisting, write down the situations that already cause friction in your cloud accounts. Most estates share a familiar set:
- A new project or service creates fresh principals and fresh grants, and nobody records the purpose.
- One role bundles far more actions than the task in front of the engineer needs.
- A grant made at a parent scope quietly applies to many child resources.
- Temporary access added to unblock a release is never taken back.
- The joiner, mover and leaver process handles the employee account while separate cloud principals keep running.
Turn each one into a test case. For example, ask the vendor to show the tool tracing a parent-scope grant down to the resources it reaches, or showing which service principals remain after the related employee has left. A shortlist built on your own scenarios is far more useful than a feature comparison.
It also helps to agree on the outcome you are buying. Least privilege is the usual goal. NIST SP 800-53 control AC-6(7) asks organisations to review privileges assigned to defined roles or classes of users, confirm each is still needed, and reassign or remove privileges where necessary. A CIEM tool should make that review practical for cloud permissions, not merely report that excess exists.
Six capabilities to evaluate in a CIEM tool
The following capabilities separate a tool that produces findings from one that supports decisions. Score each against your own test cases.
Entitlement data that keeps its scope
Ask how the tool collects cloud entitlement data and whether each record keeps its scope. A permission to read storage means something very different at the level of one resource than at the level of an entire account or organisation. Look for data that records the action, the resource and the scope together, so reviewers are never judging a grant without its boundaries.
Grants traced back to a named identity
Every permission should resolve to the identity that holds it, along with the route by which it arrived. Direct assignments are the easy part. Test whether the tool shows grants that come through roles, groups or inherited scopes, and whether it treats service accounts and workloads with the same care as people. Non-human identities often carry the broadest standing access, so a product that handles them as an afterthought will leave the riskiest grants unreviewed.
Usage data placed beside what was granted
Granted access and used access are different measures. A useful tool puts them side by side, so a reviewer can see that a principal holds administrative rights it has not exercised in a long time. Ask how usage is gathered and how recent it is. Then ask how the tool treats access that is legitimately rare, such as a quarterly job or a break-glass account, because usage alone should prompt review rather than trigger removal.
Policies collected alongside the grants
Policies set the allow and deny rules, frequently with conditions attached. If a tool reads assignments but ignores the policies around them, its picture of effective access will be incomplete. Check that policies are collected and shown next to the grants they shape, so the team can explain why an identity can or cannot reach a resource.
A clear view of the actions that need attention
A long list of findings is not a plan. Ask how the tool turns entitlement data into a short set of key actions, such as which grants to review first, which owners to contact and which permissions to narrow. The view should help a cloud lead decide what happens this week, not just describe the size of the problem.
Risk scores the team can tune
Risk ranking helps a small team focus. Ask whether the scores can be adjusted to your own priorities and at which levels they apply. A score that works only at the identity level makes it hard to see which departments, applications or managers concentrate risk. Customisable scoring lets the organisation decide what high risk means for its own estate.
Governance features that decide whether findings get fixed
Visibility is where CIEM starts, but cloud permissions only shrink when someone owns the decision and the change is followed through. During evaluation, give as much weight to these governance features as to discovery:
- Ownership: every identity and sensitive grant should map to an accountable owner, including service accounts and workloads.
- Reviews: cloud entitlements should flow into a structured access review with reviewer context, not into a spreadsheet exported from the tool.
- Attestation: owners should state that an identity remains valid and needed, with the decision recorded.
- Lifecycle: when a person moves or leaves, related cloud access should be part of the same process rather than a separate clean-up.
- Evidence: decisions and resulting changes should be recorded in a form that audit and risk teams can use.
The people who understand a permission are rarely the people who run the CIEM console. Our post on user access reviews in identity governance explains why structured review beats ad hoc clean-up.
CIEM evaluation questions to ask in a demo
Use these questions to keep a demo grounded in your own estate:
- Show one service account in our environment and every permission it holds, including inherited ones.
- Where does each grant come from, and what scope does it apply at?
- Which permissions has this identity not used, and how recent is that usage data?
- Which policies affect this identity, and where are they shown?
- Who owns this identity, and what happens when the owner leaves?
- How does a finding become a review task for the right person?
- Can we change how risk is scored for our departments and applications?
- How is an owner decision recorded, and how is the follow-up change tracked?
- How are service accounts and workloads handled compared with people?
- What does the team see first thing on a Monday: findings, or the key actions to take?
Answers given with your data, not a prepared sample, are the ones worth scoring.
Warning signs during a CIEM evaluation
Some patterns suggest that a tool will add work rather than remove it:
- Findings appear with no route to an owner or a reviewer.
- Unused permissions are recommended for removal with no check on business need or dependencies.
- Service accounts and workloads are listed but cannot be assigned owners.
- Risk scores are fixed and cannot reflect your own priorities.
- Cloud findings sit in a separate console from access reviews and lifecycle events, so every decision needs a manual hand-off.
- The demo relies on sample data and cannot show your own scopes and policies.
Together these patterns point to a tool that reports risk without helping the organisation reduce it. For broader context on least-privilege principles, NIST SP 800-207 describes zero trust tenets that include granting access with the least privileges needed to complete a task, and the CISA Zero Trust Maturity Model treats identity as one of its core pillars.
Scoring Citadel Identity360 against this checklist
Citadel Identity360 pairs CIEM capability with identity governance and administration, so cloud permissions are governed with the same owners, reviews and records as the rest of the estate.
Citadel collects the cloud entitlement record in full: the permission, its scope, the grant behind it, its usage and the governing policy. That covers the first four capabilities here from one source. On top of that record, Citadel presents an all-round view of the actions each team needs to take, so cloud and identity leads see what to work on rather than only how much exists.
Risk ranking is built in and adjustable. Teams tune how Citadel scores risk at five levels (the whole organisation, each identity, each department, each application and each manager), then direct review effort by that profile.
On the governance side, Citadel routes cloud findings into its access review module and its identity attestation flow, so each one becomes a decision by a named owner rather than an item in a report. Non-human identities stay in the same platform, each with a named owner through its full life, as our guide to governing service accounts, machine identities and AI agents together describes. Principals created outside any request process are covered in our article on service accounts made in the cloud console.
Because CIEM and IGA share one platform, lifecycle events, reviews and attestation all work from the same cloud entitlement data. For a deeper treatment of the CIEM model, the Cloud Entitlement Management CXO guide sets out the wider programme.
An illustrative pilot scope
The example below is illustrative, not a customer result. A cloud platform team picks one production account and three questions it cannot answer today: which service accounts hold administrative rights, who owns them and which of those rights have gone unused. In a Citadel pilot, the team loads entitlement data for that account, reviews the scopes, grants, usage and policies for each identity, and tunes risk scores so the production data stores rank highest.
The highest-ranked grants go to their owners through an access review. Owners keep, narrow or remove each one, and identities with no clear purpose go through attestation. At the end, the team has a short record of decisions and the key actions that followed, which is exactly the evidence a buying committee needs to judge fit.
FAQ on choosing a CIEM tool
Who should run a CIEM evaluation?
A joint team works best. Cloud platform engineers understand scopes and policies, while identity and security leaders own reviews, risk and audit evidence. Include at least one application owner who will be asked to make decisions.
Should CIEM be bought as a separate tool or as part of an IGA platform?
Buying CIEM inside an identity governance platform keeps cloud findings next to the owners, reviews and lifecycle events that resolve them, with no second console and no manual hand-off. Citadel Identity360 combines the two for that reason.
How much usage history does a CIEM tool need?
Enough to cover the work cycles in your estate. Monthly and quarterly jobs need a window that captures them, otherwise legitimate but rare access looks unused. Ask each vendor how usage is collected and how far back it reaches.
Do risk scores replace access reviews?
No. Risk scores rank where to look first; reviews and attestation record what an owner decided. Customisable scores in Citadel help direct that review effort to the identities, departments and applications that matter most.
What should a CIEM pilot prove?
That the tool can show scoped entitlements, usage and policies for your own accounts, route findings to accountable owners and record their decisions. A pilot that ends with a list of findings but no decisions has not tested governance.
Test the CIEM checklist on your own cloud account
Bring one production cloud account and the questions from this guide, and see how Citadel Identity360 shows its scopes, grants, usage and policies, ranks the risk and routes each decision to an owner. Request Demo to run the checklist live, or preview the platform in our product videos.