Scoring and exams

Create exams and auto-graded surveys

Turn closed questions into a scored quiz, decide what counts as a pass, and control how much of the result respondents see after submission.

Use scoring only when answers can be correct

Ordinary survey

Measures opinions, experiences or reported facts. There is no correct answer and Dayalogs does not score it.

Assessment

Measures knowledge against answer keys you define. Use it for quizzes, training checks, compliance tests or exams.

Satisfaction, preference, demographic and open feedback questions should remain ungraded. They can still appear in an assessment as context when ungraded questions are allowed.

Try asking your AI

“Turn this survey into a graded quiz. Use a 70% pass threshold and show respondents whether they passed.”

Choose questions Dayalogs can grade reliably

Question
Scoring
Correct when
Single choice
Exact answer
The selected option matches the answer key.
Multiple choice
Exact set
All correct options, and no incorrect options, are selected. Order does not matter.

You can give important questions more points. A three-point question then contributes three times as much as a one-point question.

Try asking your AI

“Make question 3 worth three points and leave the other graded questions worth one point.”

“Check that every graded multiple-choice question has an exact and unambiguous answer key.”

Decide what respondents see

After submission, Dayalogs can show any combination of the score, percentage, pass result and question-by-question breakdown. You can keep the detailed breakdown hidden when revealing correct answers would undermine a later attempt.

The pass rule can use a percentage or a point total. Different completion messages can then be shown to respondents who passed and those who did not.

Try asking your AI

“Show respondents their percentage and pass result, but do not show the question breakdown.”

“Create a pass ending and a different completion message for respondents who score below 70%.”

Use assessment results after collection

Completed responses include the weighted score, maximum score, percentage, pass result and per-question grading detail. These values can be exported or used to choose a completion ending.

Result
What it tells you
assessment_score
Weighted points earned.
assessment_max_score
Total points available.
assessment_percentage
Score normalized to a percentage.
assessment_passed
Whether the response met the pass rule.
assessment_breakdown
Per-question grading detail.

Answer keys remain private

Correct answers stay on the server and are not included in the survey sent to the browser before submission. Inspecting the page does not reveal the grading key.

Preview grades the test attempt without saving it as a real response, so you can validate both passing and failing paths safely.

Shuffle questions without breaking dependencies

Randomization can reduce answer sharing by position, but it must preserve the logic of the questionnaire.

  • Shuffle questions only inside blocks that have no internal conditional dependency.
  • Keep instructions or a closing item fixed when their position matters.
  • Options and matrix rows can be randomized independently.
  • If question B depends on question A, keep that sequence outside a randomized block.

Each respondent keeps the same randomized order if they resume later, and that order is included in the export.

Try asking your AI

“Shuffle the assessment questions safely, but keep any dependent sequence in its original order.”

When you need the underlying configuration

Your AI assistant can prepare the scoring configuration for you. If you author JSON directly, assessment settings live at survey level and each graded question declares its points, strategy and answer key.

View a compact JSON example
{ "assessment": { "enabled": true, "mode": "exam", "passing": { "unit": "percent", "threshold": 70 }, "results": { "show_to_respondent": true, "show_percentage": true, "show_passed": true, "show_breakdown": false } }, "questions": [{ "id": "q1", "type": "single", "title": "Which channel can share a survey?", "grading": { "mode": "auto", "points": 2, "strategy": "single_exact", "correct_option": 2 } }] }

Current limits

  • Automatic grading currently supports exact single choice and exact-set multiple choice.
  • There is no partial credit, numeric-range grading, text matching or matrix grading yet.
  • Questions inside repeat flows cannot currently be graded.
  • Answer keys are never available to respondents before submission.