Running your own mock
Turn the course's exercises into a four-round mock: 45 minutes, a written answer, a rubric score, and a retry a week later with one number changed.
On this page
A mock is the course's exercise loop run at interview pace: one situation, 45 minutes, a written answer, a binary score against the rubric, and the same situation again a week later with one number changed. This module's four exercises are one per round, so they make a loop on their own, and the earlier modules supply a second set. What a mock measures is not whether you can produce the strong answer but which rubric bullets you miss twice, because those are the ones an interviewer will find.
Build the loop from the course
| Round | First set | Second set |
|---|---|---|
| Case | The case round — a triage pilot in 45 minutes | The kickoff that asked for a chatbot |
| Deep-dive | The deep-dive — defend the eval | Normalise the export, as the coding screen |
| Customer scenario | The customer scenario — the CIO wants it on-premises | The feature request that adds six weeks |
| Behavioural | Behavioural — two answers about the slip | The champion left in week 4, told as your own story |
The second set is not easier, only unseen, so the last day can test the method on situations you have not answered before. A company whose loop has a decomposition round or a take-home maps onto these four rows as the typical loop shows: practise a take-home as the case written at home plus the deep-dive as its defence.
Run one round
Set a 45-minute clock and write in a document, not in your head. For a case, spend the first few minutes copying the situation's numbers onto the page and doing nothing else. The arithmetic comes from those lines, and a number you did not write down is one you will misquote at minute 30. For an estimate, the page holds the assumptions, the arithmetic and the number; for a critique, the defect, its size and what to do instead. Stop at 45 minutes, even mid-sentence, because the interview will.
Then open the strong answer and the rubric. Score each bullet 0 or 1: the page does what the bullet names, or it does not. "Justifies a loose precision line with the priority-lane arithmetic" is satisfied by the arithmetic, not by the phrase "precision matters less". Write one line naming the bullets you missed, in your own words. Scoring takes about 15 minutes, so a round is an hour and the loop is four.
Score it the way the debrief does
The totals on that card mislead. The case round hit 8 of 12 bullet-tries and the behavioural round 7 of 10, which reads as two middling rounds. By bullet, each has exactly one gap and nothing else wrong. A bullet hit once in two tries is noise: tired, rushed, a slip of attention. A bullet missed on both tries is a gap in the frame, and a third attempt reproduces it, because the frame is in a lesson rather than in the exercise.
So each 0 points at something to reread. The case round's sixth bullet, naming the evidence that would reverse the diagnosis, is the fourth part of saying the trade-off out loud: the choice, the constraint, the cost, and the condition that would reverse it. The behavioural round's third, the gap between knowing about a slip and telling the sponsor, is the weekly discipline in champion, blocker, executive sponsor.
Every rubric also reads aloud unchanged as an interviewer's question. A scorecard with two zeros is a list of two questions you cannot yet answer, which is the most useful thing a mock produces.
The retry: same situation, one number changed
The retry comes a week later because of what you remember. The next day you remember the strong answer's sentences; a week later you remember the frame and not the sentences, and only the frame helps in a room with a customer you have never met. A retry on the same numbers therefore measures memory. Change one number before you start, and pick one that should change the decision, not only the arithmetic.
- Case round. Make it 36 dispatchers instead of 24. Capacity becomes 360 an hour, the morning queue grows by 40 an hour instead of 160, and an emergency arriving at 10:00 waits behind about 80 requests, roughly 13 minutes, and is dispatched inside 20. A 41% miss rate can no longer be explained by queue position, so the classifier fixes nothing and the diagnosis has to move: handling, routing, or contractors arriving late.
- Deep-dive. The 26 disagreements were never adjudicated. The ceiling is now 87% raw agreement, a 96 is measuring the labellers' noise, and the first fix is adjudication, before any holdout.
- Customer scenario. The claims system runs in the insurer's own data centre, not a cloud tenant. "This building" may now be the real requirement, the in-tenant listing does not apply, and the answer moves toward on-premises with a later date that the sponsor decides.
- Behavioural. Candidate B told the sponsor in the week-3 update. B becomes a strong answer, and the critique moves to whether B should have cut scope without asking.
If the retry produces the original decision for the changed number, the first attempt was memorised, not reasoned, and the retry has done its job by showing it.
With a partner
A partner turns a written exercise into the spoken round. They read the strong answer and the rubric; you do not. They read the situation aloud, play the customer or the interviewer, and at minute 30 change one fact from the list above. Record it and score the recording against the rubric. Spoken answers lose the arithmetic first, so count the numbers you said aloud against the numbers on your written page for the same exercise. If the page had nine and the recording three, that gap is the rehearsal you need.
How many hours
Two weeks at about an hour a day covers it. Days 1–4: one round a day, an hour each. Day 5: an hour rereading the lessons behind any missed bullet. Days 8–11: each round retried exactly a week after its first try, with one number changed, which completes the scorecard and lists the gaps. Day 12: the second set as a full loop in one sitting, about four hours, spoken with a partner if you have one; day 13 is spare. That is 4 + 1 + 4 + 4 = 13 hours.
The first pass spreads the rounds out because it is for learning, and each round deserves its own reread. The full loop runs them back to back because a real loop does, and whichever round comes last is answered by someone who has already talked for three hours.
In an interview
The interviewer's notes look much like the exercise's rubric, and at some point the interviewer changes a number, a stakeholder or a constraint to see whether your answer moves with it. What is being tested is whether you rebuild the answer from the method when the situation shifts. In the role, a customer does the same thing every week without announcing it: the volume doubles, the export slips, the champion leaves.
The mistake that loses points is a memorised answer to a changed situation. The interviewer moves one number and the candidate delivers the decision that fitted the old one, fluently. The retry with one number changed is the only part of the mock that trains against it, which is why it is not optional.
Check yourself
1. Your interview is in 9 days and you have 8 hours. Plan the mock.
The retries need a week's gap, so the first tries must be on days 1–2. Day 1: the case and the deep-dive (2 hours). Day 2: the customer scenario and the behavioural (2 hours). Day 3: an hour rereading the lessons behind every missed bullet. Days 8–9: retry only the two lowest-scoring rounds, with one number changed (2 hours), and a spoken run of the case with a partner (1 hour). That is 8 hours, and it gives up the second set and two retries, not the week's gap.
2. The case round scores 5 of 6 on both tries, missing the same bullet; the deep-dive scores 4 of 6 then 5 of 6, missing different bullets each time. Where does the next hour go?
To the case round's lesson, not to either exercise. The case's single repeated miss is a gap and a third try will reproduce it; the deep-dive's misses do not repeat, so they are noise. The totals point the other way, 10 of 12 against 9 of 12, which is exactly why the card is read by bullet.
3. A partner offers to play the CIO and asks whether to read the strong answer and rubric first. Should they, and should you?
They should and you should not. The partner needs the rubric to score you and the strong answer to know when to press; you need to reach the requirement question on your own. Ask them to change one fact at minute 30, such as the claims system running in the insurer's own data centre, and score whether your proposed placement changed with it.