Data Analyst interview scorecard template
A structured scorecard for interviewing a Data Analyst: six weighted competencies, what a 1 and a 5 actually look like, and questions that surface evidence instead of opinions. Print it, or copy it into your ATS.
The Data Analyst scorecard
| Competency | Weight | Score 1 — what it looks like | Score 5 — what it looks like |
|---|---|---|---|
| Turning a business question into a measurable one Whether they can convert a vague stakeholder request into a specific question the available data can answer, which prevents most wasted analysis. | 20% | Takes the request literally and builds exactly what was asked for, without checking what decision sits behind it or whether the data supports that decision. | Recalls a request they reframed, the question they proposed instead, and the follow-up they asked the stakeholder that changed the scope of the work. |
| SQL and extraction correctness Whether the numbers they produce are actually right, since everything downstream — dashboards, decks, decisions — inherits any error made here. | 20% | Writes queries that return the right rows but cannot explain why a join duplicated them, or how they verified the count against another source. | Describes how they sanity check a result before sharing it: row counts against a known total, spot checks of individual records, and a second path to the same number. |
| Metric definition and consistency Whether the same metric means the same thing across reports, since conflicting definitions are what erodes trust in an analytics function. | 16% | Cannot say how a core metric in their last role was defined at the edges: which users counted, what time window applied, which exclusions were made. | Names a definition they wrote or corrected, the ambiguity it resolved, and how they got other teams to adopt the same version of it. |
| Reporting and dashboard design Whether their output actually gets used, which depends on showing the few numbers that change behavior rather than everything the warehouse can render. | 16% | Measures their own success by dashboards delivered, and cannot say which ones are still opened or what action any of them prompted. | Describes a dashboard they cut down or retired, the questions it was meant to answer, and how they learned which panels nobody was using. |
| Data quality and reconciliation Whether they catch broken data before a stakeholder does, since the analyst is usually the first person positioned to notice a pipeline failure. | 16% | Reports numbers exactly as the warehouse returns them, and describes data quality as a responsibility belonging to the data engineering team. | Recalls a discrepancy they chased down — a duplicated event, a changed tracking field — how they proved it, and what they told the people using the old number. |
| Stakeholder communication and recommendation Whether the analysis reaches a decision, since findings delivered without an implication leave the interpretation to whoever argues hardest. | 12% | Presents charts and lets the audience draw the conclusion, and cannot name a recommendation they made that someone disagreed with. | States the recommendation, the confidence behind it and what would change their mind, and recalls telling a stakeholder the numbers did not support the plan. |
Weights sum to 100. Agree them before the first interview, not after — adjusting weights once you have scores is how a panel rationalises a favourite.
Questions that surface evidence
Each one asks for something that already happened, in enough detail to verify. Hypotheticals reward rehearsal, not track record.
- Describe an analysis whose conclusion changed a decision. What was the finding, and how did you validate it before presenting it?
- Tell us about a time your numbers disagreed with the numbers another team produced. How did you find the source of the difference?
- What is a metric you had to define or redefine? Where were the edge cases, and who did you have to align with before it stuck?
- Describe a dashboard or report you built that was not used. How did you find that out, and what did you do about it?
- Walk us through how you check a query result before it leaves your hands. What did that routine catch most recently?
Red flags
- Cannot describe how they verify a number before sending it to a stakeholder.
- Reports only the volume of dashboards or reports produced, never a decision that followed.
- Treats data quality problems as a job for another team, with no example of catching one.
- Cannot recall a case where the data contradicted what the stakeholder expected.
How to use this scorecard
- Agree the weights with the panel before anyone interviews, and write them down.
- Every interviewer scores every competency independently, adding a note that quotes what the candidate actually said.
- Compare scores before discussing them. Discussing first anchors the panel on whoever speaks loudest.
Build a custom scorecard → · Boolean string to source a Data Analyst →
Frequently asked questions
What should I test in a data analyst interview?
A live or take-home SQL problem with a deliberate trap in it — a fan-out join, a null-heavy column — and a business question that is underspecified on purpose. The first tests correctness; the second tests whether they clarify before building.
Does a data analyst need to know Python?
It depends on the stack. SQL is non-negotiable. Python or R matters when the role includes automation, statistical work, or cleaning beyond what the warehouse handles. If the job is reporting on already-modeled tables, weight SQL, metric definition and communication far higher.
How do I compare analyst candidates who come from different industries?
Score the transferable behaviors — question framing, verification habits, metric discipline — rather than domain familiarity, and keep the same weights for everyone. Verdict scores CVs the same way, citing the exact snippet behind each dimension so a shortlist can be defended later.
When the stakes are a real hire, use evidence
These tools are quick heuristics. Verdict reads the CV against your job description and scores six dimensions with verbatim quotes as evidence — a hiring document you can defend.