Course Content
Behavioral Interview
15 sections · 30 lessons
The final week: rehearsal, a structured mock and the room
Everything before this section produced material. This week converts material into something you can deliver out loud under mild stress, which is a different skill and does not come free.
This lesson covers that conversion in the order it happens: a rehearsal protocol for the week before, a structured mock that tells you what is actually broken, and the mechanics of the room itself — pacing, reading the interviewer, and recovering when something goes wrong.
Why silent review does not transfer
Reading your story cards feels like preparation and mostly is not. Three specific reasons:
Speaking is slower and lossier than reading. A story that takes ninety seconds to read takes three and a half minutes to say, and the parts that fall out under time pressure are consistently the same: the stakes and the reflection — the two highest-scoring sections (the improved frame).
Recall order differs. On the page you see the whole card. Out loud you retrieve sequentially, and the sequence has never been rehearsed, so you narrate chronologically, which is the failure mode from Budgeting the story.
Familiarity feels like fluency. You have read the card twelve times, so it feels known. Knowing the content and being able to produce it are different, and the interview tests the second.
The rehearsal protocol
Three passes per story, in this order. Roughly fifteen minutes a story.
Pass 1 — out loud, no notes, untimed. Say the whole story to an empty room. Expect it to be bad. Note what you forgot.
Pass 2 — out loud, timed, aiming for 3:30. Use a visible timer. Most people's second pass runs five to six minutes. The cut is the exercise.
Pass 3 — recorded, then listened back. This is the one people skip and the one that changes things.
What listening back finds
Things that are invisible from the inside and obvious on tape:
| What you hear | The fix |
|---|---|
| Ninety seconds of context before anything happens | Cut to three facts |
| "We" throughout the action section | Rewrite the action as "I", once, on paper |
| No number anywhere in four minutes | Add three: scale, before, after |
| The story ends and then keeps going for forty seconds | Write the last sentence and stop there |
| Filler at every clause boundary | Slow down. Silence is shorter than it feels |
| No reflection | Add one sentence and say it every time |
The week's schedule
| Day | Time | What |
|---|---|---|
| −7 | 60 min | Three passes on your two most likely stories |
| −6 | 60 min | Three passes on two more |
| −5 | 45 min | "Tell me about yourself" until it is smooth. Script this one |
| −4 | 90 min | Mock (see the next part of this lesson) |
| −3 | 60 min | Repair the two weakest stories from the mock |
| −2 | 45 min | Follow-up drill: have someone ask "why" five times per story |
| −1 | 30 min | Read your cards once. Sleep. Do not add new material |
Running a useful mock
Day −4 is the mock. "Can you do a mock interview with me?" usually produces encouragement. A structured mock produces evidence, and the difference is entirely in what you hand your partner before you start.
Why unstructured mocks fail
Your partner wants to be helpful and has no rubric, so they report their feelings: "that was good", "you seemed confident", "maybe a bit long?". None of that is actionable, and the positive version is actively harmful because it tells you to stop working.
They also cannot see the thing that matters — whether your answer could be transcribed into rubric rows — because nobody has told them that is the test.
The setup
45 minutes, three questions, follow-ups mandatory. Not a conversation. Give your partner:
- Three questions from your weakest competencies, written down.
- The instruction to ask three follow-ups after every answer, using the moves from Surviving the follow-up drill: why that choice, what was the alternative, what went wrong.
- The scoring sheet below, printed or on screen.
- The instruction not to react warmly. Neutral face, note-taking, no encouragement.
That last point matters more than it sounds. A partner who nods and smiles trains you to expect feedback you will not get in the room.
The scoring sheet
The sheet is deliberately hostile to vibes. There is nowhere to write "seemed confident".
Reading the results
| Finding | What it means | Where to fix |
|---|---|---|
| Stakes box empty | The interviewer cannot tell it was hard | The improved frame |
| Reflection box empty | Cheapest seniority points, unclaimed | Result and reflection |
| Fewer than ~8 facts caught | Density too low; you narrated | Why structure beats fluency |
| "We" tally exceeds "I" tally | No locatable contribution | "I" versus "we" |
| Any follow-up marked "got vaguer" | The story is thinner than you thought | Surviving the follow-up drill |
| Time over 5:00 | You are eating your own follow-up time | Budgeting the story |
The "got vaguer" row is the most valuable output of the whole exercise. It identifies stories that will collapse in the room, and there is only one fix: prepare that layer properly, or choose a different story.
Choosing a partner
In descending order of usefulness: someone who interviews candidates at your target level; another engineer at or above your level; anyone at all who will follow the sheet. A non-technical partner following the sheet is genuinely useful for structure and density, because they cannot be impressed by the technical content.
In the room
Then the day itself. Mechanics: small things, all of which are recoverable, and all of which are easier if you have decided in advance what you will do.
The opening two minutes
The interviewer introduces themselves and asks for your background. Two decisions:
Ask what they want. "Would you like the two-minute version or should I go deeper on the relevant parts?" Fifteen seconds, and it prevents the most common opening error, which is a six-minute biography.
Deliver the scripted answer. See "Tell me about yourself". This is the one answer you have memorized, and it is memorized because there is only one phrasing of the question.
Pacing
Three habits, all cheap:
Signpost. "For context —", "the constraint was —", "what I did —", "the result —". You are dictating headings (How interviewers are trained and how they score).
Watch the note-taking. If they stop writing, you have left the scoreable part. Move to the result.
Offer the fork after the stakes. "That's the setup — do you want the technical side or how we got agreement?" Lets them steer, and reads as senior.
Reading interviewer signals
| Signal | Likely meaning | Response |
|---|---|---|
| "Mm-hmm" repeatedly, no writing | You are in the context section too long | Jump to the action |
| Looking at a written list | They have questions to get through | Shorten; offer the 90-second version |
| Following up on one detail three times | This is the thing they are scoring | Go deeper; do not change subject |
| "Interesting — anyway," | Your tangent lost them | Return to the main thread |
| Long silence after your answer | They are writing, or they want more | Wait five seconds. Then: "I can go deeper on any part of that" |
That last row is worth practising. The instinct to fill silence produces the forty seconds of degrading repetition that turns a good answer into an average one.
Recovering from a blank
It happens, it is survivable, and the recovery is scored.
Say it plainly and buy time. "Give me a second to pick the right example." Then take the second. Interviewers wait; five seconds of silence is normal and reads as considered.
If nothing comes, negotiate. "Nothing's coming for that exactly. Can I give you something adjacent, or would you rather come back to it at the end?" Both options are fine and both are better than a forced story.
If you lose the thread mid-story. "Sorry — where I was going with this is —" and jump to the point. Do not restart. Restarting costs ninety seconds and the interviewer already has the context.
Asking for the question again
Always allowed, never penalised, and consistently under-used:
"Can you repeat that? I want to make sure I answer what you actually asked."
Or, better, reflect it back: "So you're asking about a time I disagreed with someone more senior, specifically — not a peer?" This catches the classification error from The question taxonomy before you spend three minutes on the wrong competency.