Grade a submission

What this achieves

The grading queue holds attempts containing questions a machine cannot score. Grading them sets the final result and releases it to the learner.

Required role: Administrator or Instructor.

Steps

Go to Learning → Grading.
Find the submission, searching by student or assessment if the queue is long.
Select Grade.
Work through each question needing manual grading.
Apply a score, using the rubric where one is attached or Quick Grade where it is not.
Write Feedback for Student.
Submit the grade.

One grader at a time

Opening an attempt takes a grading lock, and the screen tells you when another grader is already reviewing it. The claim expires after 30 minutes, releasing automatically when the other grader finishes or the time runs out.

Where the lock cannot be acquired, the interface warns you to proceed with caution because other graders may be able to submit at the same time. Take that seriously — two people grading the same attempt concurrently produces one result overwriting the other.

Example: if two HC Corp UK Ltd instructors open the same submission, the second is told the attempt is locked rather than silently marking in parallel.

Auto-graded and manual questions sit side by side

An attempt shows which questions were graded automatically and which need you. Auto-graded questions display the score awarded out of the maximum; manual ones show the student’s answer and wait for your judgement.

Auto-graded scores are arithmetic against an answer key. If the key is wrong, the score is confidently wrong — so a pattern of learners failing one particular question is worth checking against the key before concluding anything about the learners.

Rubric grading and quick grading

Option	Description
Rubric	Where a rubric is attached, the scoring grid is shown and you click a level to apply its score.
Quick Grade	Correct, Partial, or Incorrect, where no rubric applies.
Points	A score entered directly, up to the question’s maximum.

Use the rubric where one exists. It is the only thing keeping your marks comparable with another grader’s on the same assessment.

AI-suggested grades are input, not verdicts

Where a question carries an expected answer, the assistant compares the learner’s response to it. That comparison is a suggestion for you to review, not a decided mark.

The model is matching text against a reference answer. It does not know what your organization accepts, whether an unusual answer is nonetheless right, or that a learner has answered a different question well. You decide the score.

Example: a learner at HC Corp UK Ltd who answers correctly in their own words may score poorly against a reference comparison. That is exactly the case a grader exists to catch.

Feedback is what the learner takes away

Feedback is optional and it is the part of grading with lasting value. A score tells someone whether they passed; feedback tells them what to do differently.

Write it to the learner, not about them.

Partial grading

An attempt can be partly graded, with auto-graded results visible and manual questions still pending. The learner sees that state and the final score updates once every question has been graded.

Finish an attempt rather than leaving it half-marked — a learner watching a provisional score is being told something that may change.

What happens next

Once graded, the result becomes available to the learner according to the assessment’s visibility settings, and the attempt leaves the queue. The completed assessment feeds a gap report against job profile requirements.