> For the complete documentation index, see [llms.txt](https://connectyai.gitbook.io/documentation/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://connectyai.gitbook.io/documentation/usage-guides-enterprise-plan/chat-features/unverified-knowledge.md).
# Unverified Knowledge
Allows moderators to remove verified status from metrics or answers, with automatic AI checks for dependencies and downstream impact to protect data trust.
The **Unverification feature** allows workspace **moderators** to remove the *verified* status from specific entities (such as metrics, measures, dimensions) or final answers within your data workspace. This is a sensitive operation because verified entities are considered **trusted knowledge** and may be reused by others across the system.
To prevent misuse or accidental trust removal, this feature is **only available to moderators**.
***
### When should I use this?
You should unverify an entity or answer when:
* A metric or measure was incorrectly verified
* Business logic has changed and the current version is outdated
* A verified entity is discovered to be inaccurate or incomplete
* You want to remove trust from a chain of linked answers that depend on faulty logic
***
### Access and Permissions
* ✅ **Available to**: Workspace moderators
* 🚫 **Restricted from**: Regular contributors or viewers
* ⚠️ **Impact**: Unverifying a key entity may cascade and affect multiple other answers or metrics that depend on it
***
### How to Use the Unverification Tool
#### **Step 0 (Pre-requisite): Select the item to unverify**
Start by selecting the entity or answer you want to mark as unverified:
* **In Metricverse**: Select the entity (e.g., metric, expression, dimension)
* **In chat history**: Select the answer that you want to unverify
Once selected, a two-step unverification process will begin.
***
#### **Step 1: Review dependent entities**
A popup will show you all entities that are **dependent on** or **linked to** the one you selected.\
You must now choose which of them should also be unverifed.
* **Example**: If a metric expression is incorrect, you might choose *not* to unverify the dimension used in it (if the dimension itself is valid).
* You can **select or deselect** entities based on which ones are truly affected.
* If there are **no dependents**, you can skip this step.
**What you’ll see:**
* A list of related entities
* Checkboxes to select which ones should lose verified status
* Tooltips showing their type, origin, and verification chain
***
#### **Step 2: Approve cascading impact**
In this step, the system scans **where else** your selected entities have been used — such as in:
* Other metrics
* Other verified answers
* Measures or dashboards
These downstream items will also lose their verified status if your selected entity is unverifed.
**At this step, you can:**
* Review the list of affected items
* Go **Back** to Step 1 to adjust your selection
* Or click **Confirm Unverify** to proceed
***
### After unverification
Once submitted:
* The selected entity/entities will lose their verified badge
* Affected answers and metrics will reflect this change in their trust status
* The system logs the action for audit and traceability
**Important**: The unverification is not reversible unless the same entity is re-reviewed and verified again.
***
### Visual Indicators
* Unverified entities will turn from the “Verified” green badge  to "Unverified" badge 
***
### FAQs
**Q: Can I unverify multiple entities at once?**\
A: Yes, if they are related through dependencies or selected during Step 1.
**Q: Will users be notified?**\
A: Currently, notifications are not automatic, but unverification is logged and visible in the audit trail.
**Q: Can I undo an unverification?**\
A: No. You must go through the verification workflow again to re-approve an entity.
***
### Best Practices
* Always unverify **at the lowest faulty level** (e.g., metric expression vs. the entire measure).
* Use **Step 1** carefully to avoid removing trust from still-valid dimensions or filters.
* Use **Step 2** to understand the **downstream impact** before confirming.
* Document why you unverifed something (in your team process or notes).
####
