Learning · August 13, 2026 · 6 min read

15 cheat sheets to prep a data analyst interview in one weekend

Data analyst interviews test recall under pressure: joins on a whiteboard, a p-value question with no warning, a metrics case with a trap inside. One-page references are the right prep format. Here are 15 free sheets mapped to the four stages of a real interview loop.

By D8A Academy

A data analyst interview is a recall exam wearing a conversation costume. Nobody lets you google the difference between RANK and DENSE_RANK mid-screen, and nobody waits while you look up what a p-value means. That is why one-page references are the right prep format: each covers one topic interviews actually probe, small enough to read, close, and reproduce from memory. Here are 15 free sheets, mapped to the four stages of a typical analyst loop, that turn a weekend into structured prep.

The 15 sheets, by interview stage:

Stage 1: the SQL screen

Almost every analyst loop starts here, live coding or a take-home. The good news: SQL screens are built from a tiny set of patterns, endlessly re-skinned.

SQL JOINs

The 4 joins you use and the two traps that duplicate rows. Interviewers love asking why revenue doubled after a join; this sheet is the answer.

The opener

SQL execution order

You write top to bottom, the engine runs in a different order. Explains the alias error question and makes every other SQL answer sound senior.

The why machine

WHERE vs HAVING

Why aggregates are not allowed in WHERE, and the one-word fix. Asked in some form in nearly every screen at every level.

Guaranteed question

Window functions

Ranking without collapsing rows, the RANK vs DENSE_RANK trio, and the top-N-per-group pattern that separates intermediate candidates from beginners.

The differentiator

Drill these in order: SQL JOINs, the execution order, WHERE vs HAVING, then window functions. If you can write the top-N-per-group pattern from a blank editor, you are ready for most screens.

SQL execution order cheat sheet: the order the engine actually runs a query in, and the alias and filter errors it explains — The execution order sheet: one diagram that answers the alias error question and makes the WHERE vs HAVING answer obvious.

Stage 2: the statistics questions

Stats questions in analyst interviews test vocabulary and judgment, not derivations. Can you explain a number out loud without misleading anyone?

Stats formulas

Center, spread, outliers, correlation, hypothesis testing, phrased the way you would say them out loud. The p-value explanation alone is worth the read.

Interview vocabulary

Mean vs median vs mode

What each average measures, when each one lies, and the one-second skew check. The salary question interviewers use to test judgment.

Classic probe

Probability distributions

Normal, binomial, Poisson, log-normal: what each models and the tell that identifies it. Enough to answer which distribution fits this with confidence.

Shape spotting

Percentage traps

Why down 50% then up 50% leaves you down 25%, percentage points vs percent, and why you never average percentages. Pure interview bait, defused.

Trick questions

The sheets: the statistics interview vocabulary, mean vs median vs mode, probability distributions, and the percentage change traps. Practice saying the answers aloud; the interview format is spoken, and fluency is half the evaluation.

Percentage change traps cheat sheet: why down 50% then up 50% is not back to even, percentage points vs percent, and the averaging trap — The percentage traps sheet: the exact trick questions interviewers reach for, defused on one page.

Why recall beats rereading

Rereading a topic five times builds recognition: it looks familiar when you see it. Interviews demand recall: producing the answer with nothing on screen. The fix costs nothing. Read a sheet, close it, and rewrite its core idea from memory. The gap between what you recognized and what you reproduced is your actual to-do list for Sunday.

Stage 3: the case and metrics round

This is where analysts are separated from query writers. You get a business scenario, signups are up, revenue is flat, what do you look at, and the interviewer watches how you think about measurement.

Misleading charts

The five tricks that make honest numbers lie, and the 5-second audit. Cases often include a deliberately bad chart to see if you catch it.

Chart critique

Vanity vs actionable metrics

The 3-question test that exposes a vanity metric, and the translation table to its actionable cousin. The exact skill the signups-are-up case is probing.

Judgment test

SMART KPIs

What separates a KPI from a metric, the SMART test, and the red flags of decorative KPIs. For the design a dashboard for X question.

Metric design

AARRR funnel

Five funnel stages, one metric per stage, where funnels actually leak. A ready-made structure for almost any product or growth case.

Case framework

Before the case round, read misleading charts, vanity vs actionable metrics, the SMART KPI test, and AARRR pirate metrics. Cases feel unstructured, but a candidate with a funnel framework and a vanity-metric detector always sounds prepared.

Vanity vs actionable metrics cheat sheet: the 3-question test that exposes a vanity metric and the translation table to its actionable version — The vanity metrics detector: the exact judgment the signups-are-up case is probing.

Stage 4: the tools round

Less glamorous, still screened: can you actually operate a spreadsheet and a notebook?

Excel formulas

The 22 formulas behind most business analysis, lookups, conditional math, spill functions, plus the three polite lies Excel tells by default.

Still everywhere

VLOOKUP vs INDEX/MATCH

The three lookups side by side and the approximate-match default that fills reports with wrong rows. Often asked by name in analyst interviews.

Named question

pandas one-liners

The 23 one-liners that carry nearly every analysis, plus the first four lines to run on any new dataset. Ideal for a live notebook exercise.

Python check

Covered in the Excel formulas that still matter, VLOOKUP, INDEX/MATCH or XLOOKUP, and the pandas cheat sheet.

pandas cheat sheet: the 23 one-liners that carry nearly every analysis, plus the first four lines to run on any new dataset — The pandas one-liners sheet, ideal prep for a live notebook exercise.

An interview never asks whether you have seen the answer. It asks whether you can produce it, right now, with someone watching.

What the loop is really measuring

The weekend plan, and what a weekend cannot do

Split the fifteen sheets across two days: SQL and statistics Saturday, cases and tools Sunday, each in read-close-reproduce cycles with a second pass on whatever you failed to recall. That is honest, high-yield prep, and it will carry you through the recall stages of the loop.

But be clear about the boundary. Every loop also contains the question no sheet answers: tell me about an analysis you did. That answer has to come from a real project, messy data, a real business question, a decision you can defend under follow-ups, and it is the part of the interview that actually produces offers. Cheat sheets sharpen recall; only applied work creates evidence. If your portfolio is thin, start there this week with our data analyst portfolio project ideas, and if you are wondering whether the effort pays, here is why an analyst portfolio is worth it. Prep the recall in a weekend. Build the proof before you need it.

Frequently asked questions

How do I prepare for a data analyst interview quickly?

Map your prep to the actual loop: a SQL screen, a statistics conversation, a metrics or case round, and a tools check. Then prep with recall, not rereading. Take one one-page reference per topic, read it, close it, and reproduce the core pattern from memory. In a weekend you can cycle the fifteen topics in this article twice, which beats skimming a 300-page book once, because interviews reward what you can produce without notes.

What SQL questions are asked in data analyst interviews?

Four themes dominate: joins, including why a join duplicates rows; the difference between WHERE and HAVING; the execution order, usually probed through an alias error; and window functions, almost always as a top-N-per-group question with ROW_NUMBER or RANK. If you can write those patterns from memory and explain why they work, you pass most SQL screens. Whole screens are built from those few patterns dressed in different table names.

What statistics do I need for a data analyst interview?

Interviewers rarely want derivations. They want working vocabulary: mean versus median and when each lies, standard deviation, correlation versus causation, what a p-value actually means, and common distribution shapes. They also love percentage traps, like why a 50% drop needs a 100% gain to recover. Being able to explain these out loud in plain language matters more than any formula, because the role is explaining numbers to non-technical stakeholders.

How long does it take to prepare for a data analyst interview?

If you already have the fundamentals, one focused weekend covers the recall work: SQL patterns, statistics vocabulary, metrics thinking, and tool syntax. What a weekend cannot build is the evidence layer, one or two real projects you can discuss in depth when the interviewer says tell me about an analysis you did. Start the portfolio work weeks earlier; use the final weekend to sharpen recall on the fifteen topics interviews actually probe.