Data analyst

What the interview listens for: How the candidate finds and corrects errors, gaps and inconsistencies before analysing data.

1. Level 1Analyses data as received; does not describe checking it.
2. Level 2Weak barRemoves obvious errors and duplicates by following a given procedure.
   • Ability to find and remove obvious errors and duplicatesMust-have
   • Awareness of the need to check data before analysing it
3. Level 3Moderate barProfiles a dataset, decides how to treat missing or inconsistent values, and records what was changed.
   • Ability to profile a dataset before working with it
   • Approach to treating missing or inconsistent values
   • Approach to recording what was changed in the data
4. Level 4Strong barBuilds repeatable cleaning steps for data others use, and explains how each choice affects the results.
   • Ability to build repeatable cleaning steps for shared data
   • Understanding of how each cleaning choice affects the results
5. Level 5Every strong indicator met, plus exceptional evidence: designs preparation across several sources, with checks that catch problems before they reach an analysis.

What the interview listens for: How the candidate retrieves, joins and summarises data held in databases.

1. Level 1Relies on exports prepared by others.
2. Level 2Weak barWrites simple queries that filter and aggregate one table.
   • Ability to write a query that filters and aggregates one tableMust-have
3. Level 3Moderate barWrites queries that join several tables correctly, and explains how the result was checked.
   • Ability to join several tables correctly
   • Approach to checking that a query's result is right
4. Level 4Strong barModels data for analysis, writes queries others reuse, and considers correctness and performance at scale.
   • Ability to model data for analysis
   • Experience writing queries others reuse
   • Awareness of correctness and performance at scale
5. Level 5Every strong indicator met, plus exceptional evidence: shapes how a team structures and reaches its data, with examples of designs that held up as needs changed.

What the interview listens for: Whether the candidate chooses suitable statistical methods and understands their limits.

1. Level 1Reports totals and averages without considering variation.
2. Level 2Weak barUses standard descriptive statistics and simple comparisons, with guidance.
   • Knowledge of standard descriptive statisticsMust-have
   • Awareness of variation beyond totals and averages
3. Level 3Moderate barChooses and applies a suitable method (a test, a regression, an experiment design) and states its assumptions.
   • Ability to choose a suitable test, regression or experiment design
   • Ability to state the assumptions a method relies on
4. Level 4Strong barHandles messy real cases (bias, confounding, small samples) and explains how the conclusion could be wrong.
   • Approach to bias, confounding and small samples
   • Ability to explain how a conclusion could be wrong
5. Level 5Every strong indicator met, plus exceptional evidence: sets the methods a team relies on, and reviews other people's analyses for statistical soundness.

What the interview listens for: How the candidate turns an analysis into a conclusion that somebody can act on.

1. Level 1Describes numbers without saying what they mean.
2. Level 2Weak barDraws a conclusion from a given analysis, with guidance on what it means for the business.
   • Ability to say what an analysis means, not only what it showsMust-have
3. Level 3Moderate barInterprets results against the business question and says how confident the conclusion is.
   • Ability to interpret results against the business question
   • Approach to stating how confident a conclusion is
4. Level 4Strong barReconciles conflicting sources, challenges the question when the data points elsewhere, and influences a decision.
   • Ability to reconcile conflicting data sources
   • Approach to challenging the question when the data points elsewhere
   • Experience influencing a decision with an interpretation
5. Level 5Every strong indicator met, plus exceptional evidence: informs senior decisions with interpretation others rely on, including when the answer was unwelcome.

What the interview listens for: How the candidate keeps data reliable over time, not only for one analysis.

1. Level 1Does not describe how the quality of data is checked.
2. Level 2Weak barReports quality problems when they are found.
   • Approach to reporting data-quality problems when foundMust-have
3. Level 3Moderate barPuts checks in place for a dataset (validation rules, reconciliation, monitoring) and follows up failures.
   • Ability to put validation, reconciliation or monitoring in place
   • Approach to following up failed quality checks
4. Level 4Strong barAgrees quality criteria with the owners of the data and monitors several sources against them.
   • Ability to agree quality criteria with data owners
   • Experience monitoring several sources against agreed criteria
5. Level 5Every strong indicator met, plus exceptional evidence: leads data-quality practice across an organisation, with a measured improvement.

What the interview listens for: How clearly the candidate communicates findings, method and limits to the people who need them.

1. Level 1Shares raw output without explanation.
2. Level 2Weak barProduces standard reports or dashboards to a given template.
   • Ability to produce standard reports or dashboardsMust-have
3. Level 3Moderate barExplains findings, method and limits in a report or presentation suited to its audience.
   • Ability to explain findings, method and limits
   • Approach to suiting a report to its audience
4. Level 4Strong barTailors reporting to different audiences, non-specialists included, and handles challenges to the findings.
   • Ability to report to non-specialist audiences
   • Approach to handling challenges to the findings
5. Level 5Every strong indicator met, plus exceptional evidence: sets reporting standards others follow, and communicates results that inform senior decisions.