The six sessions 🧭 · from the job description to the review after the last round
The running case is Ren, a constructed analyst with three years of dashboards, SQL and A/B test readouts, applying for a first data scientist role in product analytics. Ren, both job descriptions and every story are invented for teaching, and every page says so. No real employer's process is described, and nothing here is about pay.
What the interview is actually measuring
A loop as a set of measurements, what selection research says structured interviews predict and where it disagrees, what structure means from the candidate's chair, and a job description read into a prep plan.
🗂️ Session 2 · coreA story bank that covers the job
Past-behaviour questions, STAR and where it comes from, eight stories tagged against every competency they honestly evidence, trimmed to finish, and told without overselling.
🧮 Session 3 · coreThe SQL round
Read the question before typing: the grain of a join, NULLs in an average, WHERE against HAVING, date ranges, the id that is not a person. And thinking aloud while someone watches.
⏱️ Session 4 · bench nightThe practice-round bench
Six SQL questions on real orders with a strict checker that prints the first wrong row, then Ren's story bank against a job description, the one-story anti-lever and a break button.
📈 Session 5 · deeperStats, experiments and the product case
A p-value explained in one sentence, why stopping a test early is unreliable, designing an experiment out loud, choosing a metric and estimating a number in a product case.
📝 Session 6 · deeperTake-homes, your questions, and the review after
A time-boxed take-home with its assumptions written down, the questions worth asking and when, and a review after every round that feeds the bank and the drill list.
Your passport 🎫
Clear all three quiz questions in a session and it gets stamped. Six stamps means you can read a job into competencies, tell a tagged story bank, catch the common SQL traps, practise against a strict checker, reason through a statistics question and a product case, and run a take-home and a review. Stamps live in this browser only.
What this is not, so you land in the right place 🧭
| If your question is | Go here instead |
|---|---|
| How do I run structured interviews and calibrate a hiring panel? | AI + HR, Structured interviews ↗ |
| How do I build a reputation and a visible portfolio? | Personal Branding ↗ |
| How do I learn SQL, statistics or product analytics properly? | SQL ↗ · Statistics ↗ · Product Analytics ↗ |
| How do I speak with presence on the day? | Public Speaking ↗ |
| I have a data or AI interview coming and want to prepare for what it actually measures. | you are in the right place |
A note on nerves. Interviews are stressful, and parts of these sessions are about planning the workload so stress has less to feed on. Nothing here is medical or clinical advice; the body side of stress belongs to the planned move-shelf courses on the hub.
Choose your path 🗺️
Three doors into the same habit: find out what is measured, then prepare evidence for it.
What is honest here 🧾
Interview advice travels with famous numbers, and some of them do not come from where they are said to. Each claim on these pages carries its status.
| Claim | Status |
|---|---|
| Every SQL result on the bench | Really computed in your browser by SQLite on a seeded sample of the Olist e-commerce dataset (CC BY-NC-SA 4.0), and checked against an independent Python run |
| Story-bank coverage numbers | Constructed inputs, real arithmetic. Ren's stories and fits are invented; the scoring rule is printed on the widget |
| "Structured interviews predict performance better than unstructured ones" | Read at source in both the 1998 and 2022 estimates; how far the numbers should be corrected is contested, and session 1 teaches the disagreement |
| "Interviewers decide in the first four minutes" | Not what the study found. Session 1 traces it |
| STAR was invented by one firm in 1974 | A company's own claim, with no independent history found; session 2 says so |
| Keep every answer under two minutes; take-homes predict performance | No study found for either; taught as practice, never as findings |
The knowledge map 🧠
The whole path at a glance - hover a session to spotlight its concepts, click any node to jump in.