Where the questions come from
Three sources, and a stage can mix them:
- AI-generated from the position's skills and level.
- Your question bank — questions you have written or approved before.
- Specific questions you pick by hand.
See question sources for how to combine them.
How many, and how long
Set the question count and the duration. The two are related and the platform will tell you when they are incompatible — sixty questions in ten minutes is refused rather than accepted and discovered by a candidate.
A per-question timer is optional. It changes the assessment's character: it measures recall under pressure rather than reasoning, which is right for some roles and wrong for most.
Marking
Negative marking subtracts for wrong answers. It discourages guessing and it also penalises the candidate who reasons to a near-miss more than the one who leaves it blank. Off by default, deliberately.
Shuffling randomises question and option order per candidate. On by default. It makes comparing notes useless and does not change difficulty.
Passing
Set a pass threshold and whether the transition is blocking. In automatic mode the platform decides from the score; in manual mode every candidate waits for a person, which is the right choice when the stage is one input among several.
What the candidate sees afterwards
Separately configurable: their score, their score with the breakdown, or nothing until you release it. It defaults to nothing.