Build a question bank that actually prepares you
Most questions are the same few stories in different costumes
· 5 min read · Interview Preparation
Most interview question banks are procrastination with a spreadsheet attached.
You collect 120 questions. You read them. Each one you think yeah, I could answer that. You feel prepared. Then someone asks about a time you were wrong and you produce ninety seconds of unstructured autobiography, because reading a question and answering one are completely different activities.
The problem isn't the size of the bank. It's that it's built around the wrong unit.
Questions aren't the unit. Stories are.
Here's what a hundred behavioural questions actually contain: about eight distinct probes, asked in different costumes.
"Tell me about a failure," "a time you were wrong," "something you'd do differently," "a project that didn't go to plan" — that's one probe with four wordings. Prepare a hundred answers and you'll have a hundred half-rehearsed fragments. Prepare eight stories that can each be aimed in several directions, and you're covered.
So build the bank backwards. Not questions first, then answers. Stories first, then the questions each one answers.
Step 1 — Write eight stories
Pull them from your last three to five years. You want range, not volume:
- Something you built or turned around
- Something that failed, honestly told
- A conflict with a peer or a manager
- A people call — hiring, firing, underperformance, growth
- A technical or strategic judgment call with a real trade-off
- A moment of pressure — an outage, a deadline, a crisis
- Something you changed your mind about
- Something you influenced without authority
Each one needs a situation, your specific action, a quantified result, and what you'd do differently. Write them properly — full sentences, out loud at least once. Eight good stories is roughly two evenings of work and it's the highest-leverage prep you will ever do, because it's reusable across every interview you take for years.
Step 2 — Build a reuse map
Now collect questions, and as each one arrives, tag it with the story that answers it.
This single habit is what keeps the bank small and useful. The moment you write down a question, you're forced to answer: which of my eight does this? Either one of them fits — in which case you're already prepared and you move on — or none does, which is the genuinely valuable signal. A question with no story behind it is a real gap, and there are usually only two or three of them.
Your bank stops being a reading list and becomes a coverage map.
Step 3 — Tag by what's being tested, not by topic
Most people file questions by subject — "leadership questions," "technical questions." That grouping is useless, because it doesn't tell you anything about how to answer.
File by the probe underneath instead: ownership, judgment under ambiguity, failure and self-awareness, conflict, technical depth, motivation and fit. Once you can see the probe, the question stops being a surprise. "Why did you leave your last role?" and "what would your previous manager say about you?" look unrelated and are the same probe wearing different clothes.
Step 4 — Get the questions from the right places
In rough order of value:
- The job description. Every responsibility listed is a question waiting to happen. "Owns the migration roadmap" is going to become "walk me through a migration you've led." This is the richest source and almost nobody mines it.
- The company's situation. Post-layoff, post-raise, mid-replatform — each generates predictable questions about resilience, speed, or scale.
- Your interviewers' backgrounds. People ask about what they know. An infra-heavy panel probes differently than a product-heavy one.
- Your own CV. The obvious one people skip. Every gap, every short tenure, every unexplained jump is a question. Write them down and answer them before someone else asks.
- The standard set. Last, not first. It's the least specific and it's where everyone else starts.
Step 5 — Track state, not just content
Every entry needs a status, and there are only three:
- Not attempted — you've read it and nothing more
- Said out loud once — you know where the answer wobbles
- Ready — you've delivered it cleanly, inside two minutes, twice
Without a status column, a bank quietly tells you you're prepared when what you've actually done is read. Most people's banks are 100% "not attempted" and feel like 70% done.
The size rule
If your bank is bigger than about 25 questions, you're collecting rather than preparing.
Twenty-five well-chosen questions mapped onto eight stories, with a status against each, will beat a list of two hundred every time. The long list feels safer precisely because it's never finished — and something that's never finished can never be failed at.
How to actually work it
Out loud, timed, one question at a time, in random order. Reading your answers silently is the single most common mistake in interview prep: you produce the sensation of fluency without the fluency, because your brain fills in the gaps that your mouth would trip over.
Record yourself once. It's unpleasant and it's worth more than an hour of reading. You'll immediately hear the two things that need fixing — usually a story that takes four minutes and should take ninety seconds, and an answer where you never say what you did.
A question bank isn't a list of questions. It's a small set of stories, a map of what they cover, and an honest record of what you've actually said out loud.
Build it once. Maintain it across your whole career, adding a story every time something notable happens — while you still remember the numbers. The version of you interviewing in three years will be extremely grateful.
Want it generated from the real job description? labor.quest builds the question bank for a specific role and round, maps it to stories pulled from your own CV, and tracks what you've practised.
All posts