> For the complete documentation index, see [llms.txt](https://docs.openiam.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.openiam.com/administration-guide/risk/reviewer-experience.md).
# Reading risk on the reviewer screen
For reviewers — what the risk pill, the critical and high counts and the risk explanation mean, and how to work through a review queue in the time you have.
This page is for the manager or application owner who has been asked to review access. You have fifteen minutes and 300 rows. It explains what the risk marks on your screen mean and how to use them to spend your attention where it counts. It contains no settings; those belong to your administrator.
**On this page**
* [What you are looking at](#reviewing-what-you-see)
* [Where to start: the dashboard and the Certifications cards](#reviewer-where-to-start)
* [The critical and high counts on a row](#reviewer-high-risk-marker)
* [The Membership risk column](#reviewer-membership-risk-column)
* [Opening the risk explanation](#reviewer-breakdown) · [Show details](#details-fold) · [Show calculation](#reviewer-calculation)
* [The drift arrow](#reviewer-drift-arrow)
* [The Auto pill](#reviewer-auto-marker) · [The SoD marker](#reviewer-sod-marker) · [The Origin column](#origin-column)
* [List and Focus](#reviewer-priority-queue) · [Bulk decisions](#reviewer-bulk)
* [The reasons Focus gives](#reviewer-focus-reasons) · [Deciding a Focus group at once](#reviewer-focus-bulk)
* [Working through a queue](#reviewing-a-queue)
* [What risk does not tell you](#reviewing-limits)
* [The identity risk pill has been removed](#reviewer-identity-badge)
## What you are looking at
Every row you decide is one person's access to one thing, and the campaign has scored each one from 0 to 100 for risk. The score is not a verdict. It is a ranking: it says which rows deserve your attention first, and why. Scores fall into four bands, coloured green (**Low**), amber (**Medium**), orange (**High**) and red (**Critical**).
You meet risk in three places. Your first list, of people or of applications, shows a count of how many risky items are waiting inside each row. Opening a row gives you the individual items, each with its own risk pill. Clicking a pill explains, in plain words, why that item is risky.
You see all of this only in campaigns where your administrator has switched it on.
## Where to start: the dashboard and the Certifications cards
While you have items left to review in an active certification, your SelfService welcome dashboard shows a **Certifications** section below your pending approvals and your applications. It lists up to three of those certifications, one per row. Each row shows when the certification is due, and how many of your items are left to review out of your total, for example *12 of 15 left to review*. A due date that has passed reads **Overdue** in red, and the header counts the overdue certifications. In a certification launched with **Enable Beta Features** on, the row also shows how many **critical items** are still waiting for you, red when there are any.
The rows are ordered so the most urgent work comes first: overdue certifications with critical items, then other overdue ones, then those with critical items, then the rest. Within each group the one due soonest comes first. Click a row to open that certification directly. **View all** opens the full list. When you have nothing left to review, the section is not shown.
On the **Certifications** list itself, each card carries an amber **N need your attention** pill: how many of the items still waiting for you have at least one reason to look. A campaign administrator's card counts every reviewer's items instead. Sort the list by **Most needing attention** to put those certifications first.
The pill and that sort apply only to certifications launched with **Enable Beta Features** on.
## The critical and high counts on a row
On the list of people, and on the list of applications, a row can carry two small tinted counts: a red **3 critical** and an orange **7 high**. They tell you how many items **still waiting for your decision** inside that row are in those bands. A count you have already worked through disappears. A row with nothing critical or high shows no marker at all.
Hover it and it says so in full: *Still to review here: 3 at critical risk and 7 at high risk.*
Use it to choose where to start. A row with nine critical items is where your first twenty minutes should go; a row with no marker can usually be worked in bulk.
In **Focus**, a row that has items needing your attention does not show these counts beside its name. Every critical or high item is also an item that needs your attention, so the counts move into the **N need your attention** pill: rest the pointer on it to see them, together with a link to this section. In **List** the rows carry no marks at all: no attention pill, no critical and high counts, no robot and no warning triangle, in any certification. Switch to **Focus** to see them. When you certify a selection in bulk, the bar still tells you how many of the selected items need individual review.
## The Membership risk column
Open a row and each item shows a **Membership risk** pill: the risk of that one grant, for that one person. The column header explains it as the risk of this specific access assignment. In a certification launched with **Enable Beta Features** on, the pill takes the look of the Focus chips: **Critical** is filled red and **High** filled amber, while **Medium** and **Low** keep their pale tint. An item you have already decided fades behind its green or red bar, so the undecided ones stand out.
The pill combines how dangerous the access itself is with the circumstances of the grant: how it was given, how old it is, whether it has an end date, when it was last reviewed, and whether it was once removed in a review and came back. Tom Becker's `Payments-Admin` scores **98**. His `Finance-Reports-Read` scores **38**. If you have time for one of Tom's rows, the pill tells you which.
A dash means the item has not been scored. That is not a low score; it means the system had nothing to judge, so the item is yours to decide on the facts.
The column cannot be sorted directly. Use [Focus](#reviewer-priority-queue) instead.
## Opening the risk explanation
Click a pill and you get a short explanation in plain language: a headline naming the band, then at most three sentences saying what is driving it and what is fine.
For Tom's `Payments-Admin`:
> **Critical risk for this access** Several warning signs add up: it was set up directly by an administrator instead of being requested, nobody has ever reviewed it and the access itself is considered high risk. Everything else is normal: it has not expired and it was granted recently.
Read it as a story. A permission that can move money was given to a contractor outside the normal request process, and nobody has ever confirmed it. The grant is new and dated, which is why the second sentence is quiet. The decision this leads to is **revoke**, or a certification with a written justification from whoever asked for the grant.
For Maria Rossi's `Finance-Reports-Read`:
> **Medium risk for this access** It never expires. Everything else is normal: it was requested and approved the normal way, it was reviewed recently and it has not expired.
A long-standing, properly granted, recently reviewed reporting entitlement whose only mark against it is that it has no end date. **Certify.**
Three phrasings tell you how much is wrong. *Several warning signs add up* means more than one serious signal, and that the score was raised to match the worst of them rather than averaged down. A single sentence with no such preamble means one thing stands out. *Nothing unusual* means the explanation found nothing to flag at all.
If an item was never scored, the explanation says so plainly: *Not assessed yet. The risk of this access has not been calculated.*
### Show details
Under the explanation is a **Show details** link. It unfolds the full technical breakdown: every contributing factor with its weight, a bar and a score, the sentence behind each, what changed since the campaign launched, and when the score was computed. It starts folded every time you open a pill.
You do not need it to make a decision. Open it when the plain explanation does not tell you enough, or when you want to see exactly which factor moved.
### Show calculation
Inside the details is a further **Show calculation** link, which shows the arithmetic: the entitlement's own score as a base, what the circumstances of the grant added on top within their cap, an escalation line if the score was raised to the worst warning sign, and the total.
## The drift arrow
A small red up-arrow or green down-arrow beside a pill means the item **changed band since the campaign launched**. Hover it to see the before and after. When the risk went up, the plain explanation says so too: *This risk went up since the review started.*
Treat an up-arrow as a reason to open the details. Once you submit a decision, the explanation shows the risk **as recorded at the decision**, and it stops moving.
## The Auto pill
A blue **Auto** pill with a small robot means the campaign approved this item automatically at launch, because its risk was below the threshold your administrator set. Its decision is already set to certify. On a row that groups several items in **Focus**, the robot shows how many were pre-approved.
These are decided. Do not re-review them one by one. Spot-check a few, confirm the rest, and if one surprises you, treat that as feedback about the rule rather than about the item. You can still overturn one: flip its decision to revoke and submit. The fuller guidance is in [Auto-certified items](/administration-guide/risk/certification.md#cert-auto-certified-items).
## The SoD marker
A red **SoD** marker means the item is part of an unresolved separation-of-duties conflict. With **Enable Beta Features** on, it is the filled red **SoD violation** chip that Focus uses. That is a different kind of problem from a risk score: not a ranking but a conflict that has to be resolved. These items are excluded from bulk decisions and are never approved automatically.
## The Origin column
The **Origin** column shows how each grant was made: requested and approved, granted by a rule, set up directly by an administrator, and so on. It is the same fact the explanation leans on when it says an item was set up outside the normal process, shown for every row at once so you can scan for the odd one out.
## List and Focus
A **Focus | List** switch sits above your lists, and it does something slightly different depending on where you are. Earlier releases called this view **Priority Queue**. Point at the switch for a short explanation of both views and a link to this section.
On the **list of people or applications**, Focus tells you where to start. Each row shows an amber **N need your attention** count: how many of its items still waiting for you have at least one reason to look. Rest the pointer on the count for a second for a card that explains it, says how many of those items are at critical and at high risk and how many cannot be certified in bulk, and links to the documentation. The card closes when the pointer leaves; click the count to keep it open, and click anywhere else or press Escape to close it. Clicking the count does not open the row. Rows with the most such items come first, then those with the most critical items, then the most high.
On an application, one reason belongs to the row itself rather than to one item, so it appears beside the count: **Others revoked this for 40% of holders**. Whether a person stopped signing in is not shown on the person's row; it appears on each of their items once you open the row.
Clicking any column header to sort puts you back into List.
Inside a **row**, Focus sorts the items by what you should do with them, not by their risk band. It loads all of the person's or the application's items at once, up to 500, so the groups describe the whole row rather than one page:
| Group | What is in it | What to do |
| ------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------- |
| **Needs your attention** | Every item with at least one reason to look, shown as a chip | Decide each one on its own |
| **Same as last time** | Items certified in the previous campaign that covered them, with nothing changed since: the same risk band, no change since this campaign launched, the same origin and the same start and end dates | Certify them together, under your campaign's usual bulk rules |
| **Auto-certified** | Items the campaign certified at launch | Spot-check a few and confirm the rest; you can still overturn any of them |
| **Everything else** | Items that fit none of the above | Work through them last |
A line at the top counts the items in each group, for example *287 items: 9 need your attention, 41 same as last time, 237 auto-certified*. Click a count to jump to its group. If a row holds more than 500 items, the line says that it shows only the first 500; narrow the filters to see the rest in Focus.
An item without a chip reads *Nothing unusual found*. That is not a verdict that the access is safe. It means none of the signals Focus looks for fired, and the decision is still yours.
Focus opens by default when you are deciding the items. A campaign manager looking at the read-only view starts in List. Whichever you choose is remembered for you, in this browser, the next time you open a row.
In a campaign launched with Caddie — Access Review Assistant, its sparkles button sits on every item you can still decide, including auto-certified ones.
You see the switch and Focus only in campaigns launched with **Enable Beta Features** on. Without it the lists stay in List, with no switch.
## The reasons Focus gives
Each item in **Needs your attention** shows all of its chips, the most serious first, wrapping onto a second line only when the row runs out of space. **SoD violation** and **Critical risk** are filled red and **High risk** is filled amber, always ahead of the other chips, which are white. Medium risk is not a reason to look, so it never gets a chip. Rest the pointer on a chip for a second for a short card: the value behind it and the date that value was worked out, what the chip means, and a link to this table. Click the chip to keep the card open. Every chip comes from a signal the platform already records about the access; none is a recommendation.
Because the chips already say it, an item with chips does not also show its risk score, the arrow for a risk that went up, or the SoD marker. An item without chips keeps its risk score.
| Chip | It appears when |
| ---------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------- |
| **SoD violation** | The item is part of a separation-of-duties (SoD) conflict that is not yet resolved |
| **Revoked before, now back** | The access was removed in an earlier review and given back later |
| **Critical risk** or **High risk** | The item's risk is in the Critical or High band. The score itself is shown on hover |
| **Risk went up since launch** | The risk band is higher now than when the campaign launched |
| **Nobody else on the team has this** or **Few others on the team have this** | The person's peers rarely or never hold this access |
| **Never reviewed, granted 2022** | Nobody has ever certified the access, and it was granted more than a year ago |
| **No sign-in for 120 days** or **Never signed in** | The person's account is dormant |
| **Granted directly, not by role** | An administrator set the access up directly rather than through a request or a rule |
| **Others revoked this for 40% of holders** | Reviewers of other people who hold the same access revoked it at least that often in recent campaigns |
The risk-based chips appear only when your administrator's risk policy includes the factor behind them: the reinstated, peer, never-reviewed, dormant and direct-grant chips each follow one factor. The *Others revoked this* chip uses the same anonymised history as Caddie's **This entitlement across other users** block, and shows nothing below the same minimum number of reviews.
Next to the **Needs your attention** title, one chip per reason lists how many items carry it, for example **Rare on the team (2)**. All of them are on when you open the row. Click a chip to switch that reason off: items whose only reasons are switched off disappear from the group, and the title says how many are hidden. **Show all** brings them back. A hidden item is only out of sight. It does not move to another group, so it can never be certified in bulk.
## Deciding a Focus group at once
When your campaign allows bulk approval, **Same as last time** and **Everything else** each offer a **Certify** button for the whole group, and **Auto-certified** offers **Confirm**. **Needs your attention** never does. The usual bulk rules still apply: High and Critical items, and anything with an SoD marker, are left for you to decide one by one.
Every decision is still recorded against its item, with your name. A decision made with one of these buttons is recorded as a bulk decision, together with the group it came from, and the item's history says so: *Decided in bulk from the Focus group "Same as last time".*
## Bulk decisions
Select several items and the bar offers **Certify selected** and **Revoke selected**. High and Critical items, and anything with an SoD marker, cannot be certified in bulk; the bar tells you how many of your selection need individual review and leaves them for you. That is deliberate: bulk certify is for items whose risk says nobody needs to look.
## Working through a queue
1. **Start from the dashboard or the Certifications list.** Open the certification with the most items needing your attention.
2. **Start at the top of your first list.** It opens in Focus, so the people or applications with the most items needing your attention come first, each with its count.
3. **Open the first row.** It opens in Focus. Read the summary line to see how much needs you.
4. **Work Needs your attention top-down.** Hover the chips, and click the risk pill or open Caddie when the chip alone does not settle it. Write a note when you certify something critical, because someone will ask.
5. **Resolve SoD conflicts as conflicts.** An *SoD violation* chip needs a decision about the conflict, not only about the row.
6. **Take Same as last time and Everything else in bulk** after a glance for anything you recognise as wrong. Confirm the Auto-certified group the same way, with a spot-check.
7. **Escalate rather than guess.** When an explanation does not settle it, or you do not know the person or the access, delegate the row or ask, rather than deciding to make the number go away.
8. **Say when the risk was wrong.** An item that read Critical and turned out to be obviously fine, or a Low one that turned out to be dangerous, is information your administrator needs before the next campaign.
## What risk does not tell you
The score does not know your business. It cannot know that Tom's `Payments-Admin` is covered by a four-eyes control in the payments application, that his contract was just extended, or that `Finance-Reports-Read` now points at a report containing salaries. You know those things.
A high score is a question, not an accusation. "This access is correct and necessary" is a legitimate answer, and one the score can never give for you. Write it down when it is your answer, so the next reviewer and the auditor can see that a person looked and decided. A low score is not a clearance either: it means the system saw no warning signs, not that there are none. Risk decides where you look first. You decide.
## The identity risk pill has been removed
Earlier releases showed a risk pill beside each person's name, scoring that person across all of their access. It is gone from the reviewer screens.
It was removed because it answered a question a reviewer cannot act on. You decide about one grant at a time, and a person's overall score could not tell you which of their grants to look at, only that something somewhere was wrong. The [critical and high counts](#reviewer-high-risk-marker) replaced it with something you can act on: how many of the items in front of you are risky. A person's overall risk is still computed, and administrators can still read it in the Webconsole.
## Related
* [Access certification and risk](/administration-guide/risk/certification.md) — how the campaign owner set the threshold and the rule you are seeing
* [Risk management overview](/administration-guide/risk/overview.md) — what a score is and is not, for anyone
* [Risk factor reference](/administration-guide/risk/factors.md) — what each signal in the explanation measures
* [Risk policies](/administration-guide/risk/policies.md) — how the bands and the escalation rule are configured
* [Caddie — Access Review Assistant](/administration-guide/caddie.md) — the panel that puts the rest of the context beside the pill: who the person is, what the access grants, what revoking it would do
*Applies to OpenIAM 2026.10.2. Last reviewed 18 October 2026.*
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.
Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:
```
GET https://docs.openiam.com/administration-guide/risk/reviewer-experience.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.