How do I set marking criteria and points for open-ended answers in a test?

Support

Support

Last updated on November 19, 2026

You can write the marking criteria for an open-ended answer yourself or ask AI to draft them. If you have a good example of a correct answer to the test question, give it to AI. Check the criteria it suggests and edit them if needed.

Where can I find the fields for marking a test answer?

The ‘Points’ and ‘Content assessment criteria’ fields are below the question text. They appear for ‘Free answer’, ‘Audio answer’, ‘Video answer’ and ‘Upload file’ questions. For audio and video answers, you can also turn on sound assessment to see the ‘Criteria for grading sound’ field. For a video answer, you can separately turn on assessment of visible actions and fill in ‘Criteria for grading visible actions’.

How do I set the points and write my own criteria?

Each part of the answer has its own ‘Points’ field. Enter the most a learner can earn for that part. The field starts at 1 and accepts any whole number from 1 to 100. Give more points to the skill that matters most for the task.

Suppose you’re training sales advisers and ask a learner to present a pair of trainers to a customer. In ‘Content assessment criteria’, write what you want to hear from the adviser. For example, “Explains which surfaces the trainers suit and how the flexible sole benefits the customer.” If a full explanation is worth 4 points, enter 4 in the ‘Points’ field beside it. There’s no need to repeat the number in the criteria.

When should I mark sound or actions on camera?

Use content criteria alone if you only want AI to mark the substance of the answer. If delivery matters, use ‘Criteria for grading sound’ to describe what to listen for, such as intonation, confidence in the learner’s voice or where they pause. If you also want to mark what the learner shows on camera, fill in ‘Criteria for grading visible actions’.

In the example with the trainers, you might want to mark how the adviser speaks and demonstrates the product as well as what they say about it.

  • How the presentation sounds. For an audio or video answer, select ‘Sound affects the grade (optional)’. A ‘Criteria for grading sound’ field appears below, with its own ‘Points’ field. You could write “The adviser speaks clearly and pauses after the explanation to give the customer time to ask a question” and allow 2 points for delivery.

  • What the customer sees. For a video answer, select ‘Assess visible actions (optional)’. In ‘Criteria for grading visible actions’, you could write “The adviser bends the sole and turns the trainer towards the customer to show where it flexes.” Enter 4 in the ‘Points’ field beside it for the demonstration.

For video answers, you can select either option or both. Analysing sound or video frames can use substantially more AI tokens.

How do the marks add up?

AI marks each part from zero up to its maximum, then adds the marks together. It can award a fractional mark if the learner meets some of the requirements.

In our example, the adviser explains the benefits of the trainers (4 out of 4) and speaks clearly, pausing for a question (2 out of 2). They bend the sole, but the customer can’t see where it flexes. If AI awards 1.5 out of 4 for the demonstration, the total is 7.5 out of 10. AI explains the marks in its comment.

How do I use AI to generate marking criteria from a sample answer?

Prepare an example of a correct answer to the question.

Question type What to use as the example Format and limits
‘Free answer’ A written answer. For example, write what an adviser might say to explain the benefits of the trainers to a customer. Up to 30,000 characters.
‘Audio answer’ A recording of a spoken answer. For example, record an adviser presenting the trainers to a customer. MP3, WAV, MP4 with audio, OGG or WebM, up to 500 MB.
‘Video answer’ A video of someone carrying out the task. If you’re marking a product demonstration, make sure the recording shows how the learner should present the product. MP4, WebM or MOV, up to 500 MB and up to 10 minutes long.
‘Upload file’ An example of the finished document you expect from the learner, such as a product presentation for a customer. PDF, DOCX, PPTX, XLSX or TXT, up to 500 MB.
  1. If you want criteria for sound or actions on camera, tick the relevant boxes. Below the criteria fields, select ‘Generate grading criteria from an example’.

  2. For a ‘Free answer’ question, type or paste the text into ‘Example of a correct response’. For the other question types, choose ‘Select file’. Wait for the upload and file check to finish. The filename will appear beside the button.

  3. Select ‘Generate’. When you see ‘AI is preparing draft instructions’, you can close the tab or go to another page. AI will carry on preparing the draft. Return to the question after a few minutes to check the result.

How do I check the criteria and save the question?

The draft appears in the content criteria field and in the fields for any options you selected. If you’ve already written some criteria, AI adds its suggestions below yours, leaving a blank line between them. It keeps the points you set. Read through the combined text, remove any repetition and keep only what the task asks the learner to do. If AI couldn’t draft criteria for one part, you’ll see a message beneath that field. Write the missing criteria there.

Once the criteria are ready, check the points and select ‘Save’ below the list of questions. If you want AI to mark learners’ answers, open the test’s ‘Settings’ tab and choose the marking method.

What if a learner has already started the test?

That attempt keeps the criteria and points that applied when the learner started. If you change the wording or the points and save the question, the changes apply to subsequent attempts.

How do I reopen a finished draft?

Open the same question using the same account within 30 days. The date beside the generation button tells you how long the draft is available. If you have since changed the settings, select ‘Restore draft with its settings’. This recovers the existing text without generating it again or using more AI tokens. If you have changed the question wording or type, prepare new criteria for the revised task.

What happens to the example file?

The system removes files uploaded more than a day ago once they are no longer needed for generation. It keeps the draft separately, so you can still reopen it after the file has been removed.

What if the upload or generation was interrupted?

If you refreshed the page while the file was uploading, select the same file again in the same tab. If you closed the tab, reopen the question and check whether you need to attach the example again.

If the upload finished but generation stopped with a message about insufficient AI credits, top up the balance. Then return to the question and select ‘Generate’ again.