How marking works
Each answer is marked on 4 areas, scored 0 to 4 each. Every mark points at a line you said. Marks are practice feedback, never a hiring verdict.
The areas
- Ownership Did you say what you decided, and why, in the first person?
- Result Did you end on an outcome, with a number if you have one?
- Follow-up Did you answer the probe with a new, specific detail?
- Clarity Did you lead with the point and stay in order?
System design mocks are marked on 5 areas instead, one for each scored phase.
Marks that quote you
Marks come as Keep and Fix. Each one quotes you exactly. A mark that cannot quote you is not shown.
Not observed is not zero
If you did not show an area, it is marked not observed. It is left out of the total, so a short answer is never punished with a zero.
What never changes a score
Pace, accent, pauses and filler words never change a score. Marks look only at what you said.
Try lines
A Try line rewrites the line you said, using only facts you gave. Where a fact is missing it leaves a blank, such as [your number], so it never makes one up.
An example
Example
One answer, the follow-up it drew, its marks, and a second take of the same answer.
AnswerSo we migrated the payments queue, and we got latency way down. I set up a weekly check-in with the payments team so nobody was surprised. It unblocked the launch.
You said "we migrated the payments queue". Which decision in that migration was yours?
You saidwe got latency way down
Fix "Way down" from what, to what? A number makes the result checkable.
Try Latency went from [your starting number] to [your ending number].
You saidIt unblocked the launch
Keep You tied the result to a business reason.
Take 1
we got latency way down
Take 2
I chose a two-week dual-write over a cutover, so the payments queue never stopped. p99 went from 1.8s to 240ms, which unblocked the launch.
You saidI chose a two-week dual-write over a cutover
Keep The decision is yours now, and the result has numbers.
What the AI decides
- The AI chooses practice questions and follow-up questions.
- The AI writes marks from your transcript and anchors each mark to quotes you can check.
- The AI writes Try lines to help you practice a clearer version.
What code decides
- Code fixes plan prices and lengths.
- Code counts your mocks left and applies dates and refund eligibility.
- Code calculates rubric totals with fixed scoring math.