Skip to content

Reviewer

The Reviewer role is for quality assurance - people who inspect AI-generated scores, validate results, and override grades when the AI evaluation needs adjustment.

What can a Reviewer do?

  • View all assignments and their statuses
  • Approve or return course completions on courses that require sign-off
  • Inspect score reports with per-criterion breakdowns
  • Override grades when the AI score doesn't reflect actual performance
  • Add notes and feedback to score reports
  • View the dashboard with workspace metrics
  • View workspace Analytics, including the activity feed
  • Post and reply in the workspace Forum - ask a question, follow a thread, mark your own thread resolved

Approving course completions

Some courses are set up so that a learner finishing them is not the end of the story: a reviewer has to sign off first. On those courses, a finished course waits for you before it counts as complete, and before any certificate is released.

Courses only work this way when a trainer has switched on Require reviewer approval for that course. Most courses do not, and nothing waits for you on those.

Find what is waiting

  1. Go to Analytics in the sidebar.
  2. Open the Reviews section. It sits next to Evidence.
  3. Completions waiting on a decision are listed oldest first, with a count beside the heading.

Each row shows the learner, the course, when they finished, and their score. View debrief opens their full score report if you want the detail before deciding.

Approve or return

Action What happens
Approve The course counts as complete. The learner is emailed, and if the course issues a certificate, it becomes available for them to download.
Return to learner The course opens up again so they can retake it. The learner is emailed your note explaining what to change.

Returning a completion requires a note - the learner sees exactly what you wrote, so be specific about what needs to change.

If the learner has already used all their retries, returning the completion grants them one more, so they are never left with a course they can reopen but cannot start. The row tells you when this applies.

Approving does not always produce a certificate

A course can require review without issuing a certificate at all, in which case approving simply confirms the completion. And on a course that does issue one, the learner still has to meet its pass mark. Where either applies, the row tells you before you decide.

Waiting completions hold everything up

Until you approve, the completion does not count towards completion figures and does not unlock the next step of a learning path. A queue left unattended leaves learners stuck, so check it regularly.

Typical workflow

1. Check assignments

  1. Navigate to Assignments in the sidebar
  2. Review the assignment list - look for assignments with status completed
  3. Click into a completed assignment to open its Debrief

2. Review a score report

The Debrief shows:

Field Description
Overall Score Aggregate score (typically out of 100)
Breakdown Per-criterion scores based on the rubric
Feedback AI-generated qualitative feedback
Recommendations Suggested next steps for the learner

3. Override a score (if needed)

If you disagree with the AI's evaluation, the Debrief gives you three free-text fields to override its output:

  1. Open the Debrief for the assignment.
  2. Edit any combination of:
    • Overall Score Override - replace the aggregate score the AI produced.
    • Feedback Override - replace the qualitative feedback the AI wrote.
    • Next Step Recommendation - replace the recommended next step.
  3. Click Save Reviewer Override.

The original AI score and feedback stay on record alongside your override, so the audit trail shows both versions.

When to override

Common reasons to override an AI score:

  • The AI misinterpreted a learner's intent
  • Cultural or contextual nuance the AI missed
  • Technical issues during the session affected scoring

Best practices

Spot-check, don't review everything

Focus on edge cases - very high or very low scores, first-time learners, and flagged sessions. Trust the AI for routine evaluations.

Document your overrides

Always add a note when overriding a score. This creates an audit trail and helps improve future AI scoring accuracy.