Data interview questions

Role-specific Data interview questions with what each question is testing and how to prepare without memorising scripts.

Data roles at startups can combine analysis, communication and technical work. An interview can show whether you can turn messy data into a useful decision.

These are preparation questions, not a claim that every UK startup asks the same script. Strong interview processes vary by company, role and seniority.

What interviewers may be trying to understand

  • data preparation
  • analysis and model choice
  • communication
  • business impact
  • ethics and privacy
  • software and data tooling
01Tell me about analysis that changed a business decision.
What it tests

Impact.

How to prepare

Explain the decision, data, method, uncertainty and what happened next.

02How would you investigate a fall in conversion?
What it tests

Analytical structure.

How to prepare

Define the metric, validate data quality, segment, form hypotheses and test them.

03What do you do when the data is incomplete?
What it tests

Judgement under uncertainty.

How to prepare

Discuss missingness, bias, proxies and communicating confidence.

04Explain a model you built to a non-technical audience.
What it tests

Communication.

How to prepare

Focus on decision consequences and limitations.

05How would you check whether this metric is trustworthy?
What it tests

Data quality.

How to prepare

Cover definitions, lineage, joins, missing data, instrumentation and sanity checks.

06When is a simple model better than a complex one?
What it tests

Pragmatism.

How to prepare

Discuss interpretability, data volume, maintenance and marginal performance.

07Tell me about a data mistake you made.
What it tests

Learning.

How to prepare

Explain detection, impact and prevention.

08How would you design an experiment here?
What it tests

Causal reasoning.

How to prepare

Define hypothesis, outcome metric, unit, sample and confounders.

09What ethical or privacy concern would you raise about this dataset?
What it tests

Data responsibility.

How to prepare

Identify concrete harm and mitigation.

10How would you productionise an analysis used every week?
What it tests

Repeatability.

How to prepare

Move from notebook or manual work to tested pipelines, monitoring and ownership.

How to prepare well

Prepare examples that show business impact as well as technical skill. Be explicit about uncertainty. If a case gives you a dataset, validate definitions before analysing.

A useful answer structure

For behavioural examples, STAR, Situation, Task, Action, Result, is a useful starting point. Keep the context short and spend most of the answer on your own decisions and actions.

For hypothetical questions, explain assumptions and trade-offs rather than jumping to one “correct” answer.