You will see a mix of technical, business, and behavioral prompts. Prepare concise, decision-oriented answers that make assumptions explicit and finish with recommended actions.

Technical / SQL & Statistics

Expect hands-on problem solving that tests query fluency and statistical reasoning.

• Write a SQL query to compute 7/14/28-day retention by acquisition channel and flag significant outliers.
• Given an A/B test with uneven sample sizes and a noisy metric, how do you assess lift and risk?
• What are Type I and Type II errors, and how do you manage them in practice?
• How would you detect and address Simpson’s paradox in a funnel analysis?
• When is a non-experimental method preferable to an A/B test?

Experimentation & Metrics Design

We test your ability to design interpretable tests and define success rigorously.

• Design an experiment for a new onboarding flow: hypotheses, metrics, guardrails, and sample size.
• Pick a north-star metric for user engagement and define supporting input metrics.
• How do you handle peeking and decide on sequential testing?
• What variance reduction techniques do you consider and when?
• How do you respond when the treatment improves clicks but hurts revenue?

Product Sense & Business Acumen

Show how you connect insights to strategy and prioritization.

• Prioritize three growth ideas with the same engineering cost but different uncertainty profiles.
• Size the opportunity for upselling an existing customer segment; what inputs do you need?
• Propose a framework for reducing churn in the SMB segment.
• Which metrics would you review weekly for a marketplace’s supply-demand health?
• Recommend ship/iterate/rollback for a feature with mixed signals across segments.

Problem-Solving / Case Studies

Demonstrate structure under ambiguity and a bias for actionable next steps.

• Diagnostic case: Revenue dropped 7% week-over-week—structure your approach and top three analyses.
• New market entry: What data would you gather, and how do you build a quick sizing model?
• Resource allocation: You have capacity for one of three initiatives—choose and defend with data.
• Data gap: How do you proceed when a critical event is missing or unreliable?
• Time-boxed decision: Deliver a recommendation in 24 hours with incomplete data—what’s your plan?

Behavioral / Leadership & Communication

We look for ownership, clarity, and influence without authority.

• Tell me about a time your analysis was challenged—how did you respond and what changed?
• Describe a high-stakes decision you influenced across skeptical stakeholders.
• Share a failure and what you changed in your process as a result.
• How do you ensure data quality and trust in dashboards you own?
• Give an example of simplifying a complex analysis for executives.

This visual outlines the typical flow from recruiter screen through onsite assessments and team conversations. Use it to pace your preparation: pair SQL/statistics practice with structured case drills and communication reps. Plan buffers between stages to synthesize learnings, sharpen weak spots, and prepare targeted questions for stakeholders.

Deep Dive into Evaluation Areas

Analytical Rigor: SQL, Statistics, and Experimentation

Analytical depth is a core differentiator at Acumen. We assess your ability to query large datasets, select appropriate statistical methods, and design interpretable experiments. Your interviewer will prioritize clarity in problem setup, efficiency in execution, and discipline in interpreting results.

• Be ready to go over:

  • SQL fundamentals and performance: Joins, window functions, CTEs, aggregations, subqueries, and query optimization basics.
  • Descriptive and inferential statistics: Distributions, confidence intervals, hypothesis testing, p-values, Type I/II errors, power.
  • Experimentation: A/B test design, metrics selection (guardrails vs. north-star), sample sizing, bias control, and interpreting lift.
  • Advanced concepts (less common): CUPED/variance reduction, sequential testing, multi-armed bandits, propensity scoring, quasi-experiments.

• Example questions or scenarios:

  • “Write a SQL query to compute cohort retention and identify segments with statistically significant deltas.”
  • “Design an experiment to test a new onboarding variant; walk through hypotheses, metrics, sample size, and expected risks.”
  • “You observe a 1.2% lift with p=0.06—ship or hold? Justify your decision using business impact and risk.”

Business Strategy & Product Sense

Strong candidates connect data to outcomes. We evaluate how you frame opportunities, size markets, prioritize trade-offs, and define success metrics. You should translate insights into business language and recommend actions that balance risk, cost, and time-to-value.

• Be ready to go over:

  • Opportunity sizing and ROI: TAM/SAM/SOM approximations, cost-benefit models, sensitivity analysis.
  • Metrics architecture: North-star vs. input metrics, counter-metrics, funnel design, and leading vs. lagging indicators.
  • Prioritization: Impact vs. effort, strategic alignment, and sequencing bets across horizons.
  • Advanced concepts (less common): Flywheel effects, network externalities, LTV/CAC modeling with retention curves.

• Example questions or scenarios:

  • “Prioritize three initiatives for user activation given limited engineering capacity—what do you ship first and why?”
  • “Define a success framework for a new self-serve feature, including guardrails and leading indicators.”
  • “Size the revenue impact of reducing churn by 10% in the SMB segment; state assumptions and run sensitivities.”

Problem Framing & Case Structuring

We assess whether you impose structure on ambiguity and navigate from question to decision. You should create a clear roadmap from context to choice, articulate assumptions, and identify the highest-leverage analyses.

• Be ready to go over:

  • MECE structuring: Laying out drivers, hypotheses, and decision criteria.
  • Root-cause analysis: Funnel breakouts, segmentation, and correlation vs. causation checks.
  • Prioritized analysis plans: What you will do first, what you’ll defer, and why.
  • Advanced concepts (less common): Decision trees, Bayesian updates, expected value frameworks with uncertainty.

• Example questions or scenarios:

  • “Weekly revenue is down 7%—diagnose the drivers and propose next steps in 20 minutes.”
  • “You have incomplete data and a 24-hour deadline—how do you deliver a defensible recommendation?”
  • “Walk through your approach to validating a hypothesis when the ideal experiment isn’t possible.”

Data Communication & Stakeholder Influence

Insight only matters if it changes decisions. We test how you tailor messages, build alignment, and handle pushback. The best candidates bring clarity—framing trade-offs, calling out risks, and closing meetings with explicit decisions and owners.

• Be ready to go over:

  • Executive-ready narratives: Situation, key insight, decision, and impact in under two minutes.
  • Stakeholder management: Mapping incentives, pre-reads, and alignment loops.
  • Visualization and storytelling: Choosing the right chart, making outliers and deltas obvious, annotating for action.
  • Advanced concepts (less common): Influence without authority, pre-mortems, and decision journals.

• Example questions or scenarios:

  • “Present your analysis of an experiment to a skeptical PM and an engineering lead—anticipate objections.”
  • “Translate a complex statistical conclusion into a one-slide executive update.”
  • “Resolve a metric definition conflict between Marketing and Product—what framework do you use?”

Technical Execution & Tooling

You must be efficient with the tools that power modern analytics. We evaluate fluency and judgment: can you choose the right tool, automate the boring parts, and ensure reproducibility?

• Be ready to go over:

  • Core tools: SQL (warehouse-native), spreadsheets for quick modeling, BI dashboards (Looker/Tableau), version control for queries.
  • Scripting for analysis: Python or R for statistics, visualization, and notebooks; when to use vs. when not to.
  • Data quality: Event instrumentation basics, data validation, anomaly detection, and documentation.
  • Advanced concepts (less common): dbt modeling basics, airflow/orchestration concepts, feature flag frameworks.

• Example questions or scenarios:

  • “Refactor a SQL query for readability and performance; explain trade-offs.”
  • “Build a minimal KPI dashboard spec—define entities, grains, and refresh cadence.”
  • “Audit an event taxonomy for a new feature—what do you log and why?”