How to prepare for a data analyst interview in one week
A week is enough to close one or two gaps and rehearse every stage once. It is not enough to learn SQL from scratch — if that is where you are, take the easy problems and stay on SQL all week.
Day 1 — find the gaps
about 1.5 hours
- Paste the posting into Prepare for a job: you get the topics its interviews will likely cover, a plan in that order and a readiness report that marks each requirement Ready, Almost or Gap.
- No posting yet? Offer builds a general plan instead.
- Do the warm-up in the SQL track and note which problems felt slow — that is your list for Days 2 and 3.
Day 2 — SQL: grouping, joins, duplicates
about 2 hours
- Work through the first sections of SQL 30: grouping and ratios, then JOINs, NULLs and duplicates. Hints are fine; the point is a second attempt without them.
- Before each query, name the grain: what one row of the result stands for.
- Fix a churn query: finding someone else’s bug trains the check you will run on your own code.
Day 3 — SQL: dates, top-N, windows, retention
about 2–3 hours
- Next sections of SQL 30: dates and time zones, then top-N and window functions. Finish with weekly cohort retention, usually the hardest pattern on this list.
- Stuck on a pattern? Open its cheat sheet card, close it, and write the query again from memory.
Day 4 — a work week at Carrot
about 2–3 hours
- Play the work week: five game days at Carrot, a practice grocery delivery company — emails with deadlines, SQL on its database and an A/B test that needs a decision.
- Treat it as the case round: every number you send is checked by code, and the feedback compares you with what a senior analyst would do.
Day 5 — metrics, A/B and the product case
about 1.5 hours
- Two cases without an interviewer: write your plan — what to measure, which cuts, which checks, what you would decide — then compare it with the strong candidate’s walkthrough.
- For every metric in the posting, say the formula and what moves it. The cheat sheet has the cards.
- Design one A/B test out loud: primary metric, two guardrail metrics (the ones that must not get worse), how long it runs, what you do if there is no difference.
Day 6 — a mock interview, then Python
about 1.5 hours
- Take one mock interview on the stage you fear most — SQL, product case, A/B, Python or behavioral. A free account includes one; read the debrief and work on what it flags.
- Python or Pandas in the posting? Do the warm-up in the Pandas track or a few Python problems.
Day 7 — your stories, your questions, rest
about 1 hour
- Three stories, two minutes each: a project with a number at the end, a disagreement, a mistake caught in time. Say them out loud, not in your head.
- Write three questions about the team’s metrics and data. Check the readiness report: a Gap on a topic the posting names is your last hour; everything else waits.
- Stop early. A rested head beats one more problem.
Only three evenings? Keep Day 1, Day 4 and Day 6: the gaps, the work week and the mock interview.
By Day 7 you can say yes to
Two or more unchecked? Spend the last evening on those, not on new topics.
Paste your posting
Offer shows the topics its interviews will likely cover and builds a plan for that job. Free, no card.