Data Analyst Interview Questions: What UK Employers Actually Ask in 2026
UK analyst interviews test three things, and since 2025 a fourth. Preparing only for the SQL is why capable candidates get rejected.
UK data analyst interviews test three things and, since about 2025, a fourth. Preparing for the SQL alone is why capable candidates get rejected.
Stage 1: The technical test
What SQL questions come up in a UK data analyst interview?
Expect a query that joins two or three tables to answer a business question, aggregation with GROUP BY and HAVING, and a window function problem such as the latest record per customer or month-on-month change. You may also be asked to spot a bug in a query, or to explain when INNER and LEFT JOIN give different answers.
How would you know if this result is wrong?
This is the question that separates candidates, and most have no answer. Prepare one: check row counts before and after each join, reconcile totals to a figure you already trust, and sense-check whether the size of the number is plausible. Interviewers want to see that you verify a result before you report it.
Usually SQL, sometimes Excel, occasionally a take-home task.
What comes up most:
- Write a query joining two or three tables to answer a business question
- Aggregate with GROUP BY and filter with HAVING
- A window function problem: latest record per customer, ranking within a group, month-on-month change
- Spot the bug in a query or formula
- Explain the difference between INNER and LEFT JOIN, and when the choice changes the answer
The question that separates people: "How would you know if this result is wrong?" Most candidates have no answer. Prepare one: check row counts before and after joins, reconcile totals to a known figure, sense-check the magnitude.
Excel variants: build a pivot from raw data, XLOOKUP across sheets, find the error in a broken formula.
If a live coding test is involved, narrate as you go and clarify the question before writing anything. Interviewers frequently leave the brief deliberately ambiguous to see whether you ask.
Stage 2: The case study
Sales dropped 15% last quarter. How would you investigate?
There is no single correct answer; interviewers are assessing structure. Clarify what the drop is measured against, verify it is real rather than a data issue, break it down by volume, price, segment and region, form two or three testable hypotheses, then recommend a first step. Verifying the data is the step most candidates skip.
The classic form: "Sales dropped 15% last quarter. How would you investigate?"
There is no correct answer. They are assessing structure. A workable framework:
- Clarify. Fifteen percent against what: last quarter, or the same quarter last year? Which products, regions, channels?
- Verify. Is the drop real, or a data issue? A changed definition, a broken feed and a missing region all look exactly like a sales drop.
- Decompose. Volume or price? Which segments? All regions or one? New or returning customers?
- Hypothesise. Two or three plausible causes, each with the data you would use to test it.
- Recommend. What you would do first, and what you would need to be confident.
Step 2 is the one candidates skip and interviewers are specifically listening for.
Stage 3: Behavioural questions
Tell me about a time you found an error in your own work.
This comes up more often than people expect, and the wrong answer is claiming you have never made one. Describe a real mistake in situation-action-result form: how you caught it, how you put it right, and what you changed in the way you work so that it would not happen again.
Standard, but with analyst-specific angles:
- Tell me about a time you found an error in your own work
- A time you disagreed with a stakeholder about what the data showed
- How you handled a request you thought was the wrong question
- A time you had to explain something technical to a non-technical audience
- How you prioritise when three people want something the same day
Prepare three stories in situation-action-result form and adapt them. The error question is asked more often than people expect and the wrong answer is claiming you have never made one. The right answer is a real mistake, how you caught it, and what you changed.
Stage 4: The AI questions, new and increasingly common
AI now appears in 39.6% of UK data analyst adverts, and interview questions have followed:
Do you use AI tools in your work, and how?
Say yes, and be specific. Denying it reads as either dishonest or out of touch. Describe your actual workflow: which tasks you use AI for, and how you check what it produces before anything reaches a stakeholder. Interviewers are listening for honest, practical use rather than enthusiasm or avoidance.
How do you check AI-generated output?
This is the real question behind the AI topic. Answer concretely: verify the join grain, check row counts, reconcile totals against a known figure, and never ship a number you cannot derive yourself. Candidates who answer this well stand out, because most either avoid the topic or over-claim.
Where would you not use AI?
Good answers name three situations: anything involving sensitive data, anything where you cannot verify the result, and any case where you would be accountable for a number you do not understand. Showing that you know where the limits are signals judgement, which is what the interviewer is really testing.
Candidates who handle this well stand out considerably, because most either avoid the topic or over-claim.
The portfolio walkthrough
If you are a career changer, expect this and expect it to take a while. See the preparation notes in our portfolio guide. The essentials: four minutes without a screen, and honest answers about what you got wrong first.
A two-week preparation plan
Week 1: Daily timed SQL practice, concentrating on joins, aggregation and window functions. Write out your three behavioural stories. Rehearse the portfolio walkthrough aloud, recording yourself once.
Week 2: Work through three case studies out loud using the five-step framework. Research each employer: their sector, their data maturity, what they would plausibly be measuring. Prepare your AI answer. Prepare two questions to ask them that show you have thought about their business, not questions about the role's day-to-day.
The thing most candidates get wrong
Treating the interview as a test of knowledge rather than a demonstration of how you think.
Interviewers hire analysts they will have to trust with numbers. Everything they ask is a proxy for one question: will this person tell me when something is wrong, or will they hand me a confident answer they have not checked?
Answer that question well and the SQL syntax matters much less than you think.
Frequently asked questions
How long are UK data analyst interviews? Typically two to three stages: a screening call, a technical test or take-home, and a final interview combining case study and behavioural questions.
Should I do the take-home assignment? Usually yes if it is scoped to two or three hours. Anything larger is a reasonable thing to push back on politely.
What should I ask them? Questions about data maturity are excellent: where does the data live, who owns quality, how are analytics requests prioritised. They signal experience and tell you whether the job is real analysis or report maintenance.
Mock interviews and portfolio presentation are built into Uptrail's programme rather than treated as optional extras, because the format itself is what most career changers find hardest the first time.
Sources: ITJobsWatch, Data Analyst skills co-occurrence data, 6 months to 1 September 2026.
Read next
Rehearse the format before it counts
Mock interviews and portfolio presentation are built into Uptrail's programmes, alongside dedicated career coaching, because the format is what career changers find hardest first time.
Explore the Data Career ProgrammeBook a free consultation
