Marketing Analytics Interview Guide: Case Studies, Metrics & Attribution

The core loop across companies

1. SQL / spreadsheet round (funnel, cohort, retention queries).
2. Case study — usually attribution, incrementality, or a channel-mix decision.
3. Metrics deep-dive — how you'd define, instrument, and diagnose a KPI.
4. Behavioural — stakeholder management with marketing, growth, and finance.

Metric definitions you must have crisp

• CAC, blended CAC, paid CAC — and how you'd defend each definition.
• LTV (contribution-margin-based, not revenue-based) and the LTV:CAC ratio's honest use.
• ROAS vs MER — and why finance teams increasingly prefer MER.
• Retention curves — DAU/MAU, N-day retention, and quick-ratio for growth.
• Incrementality: what does 'incremental' actually mean vs 'attributed'?

Attribution — the question you will get asked

Expect: 'How would you attribute a conversion that touched Meta ads, a Google search, an email, and a direct visit?' Strong answer: describe the trade-offs of last-touch, first-touch, linear, time-decay, and data-driven; explain why every touch-based model over-credits low-funnel; then recommend geo or holdout-based incrementality as the honest ground truth. Land the answer with 'and here's what I'd implement first if we had 6 weeks'.

Case-study patterns that come up

1. 'ROAS on Meta dropped 30% last week — diagnose.' (segment: campaign, creative, audience, iOS vs Android, competitor bidding).
2. 'Should we double our Google Ads budget?' (marginal ROAS, saturation curves, incrementality test design).
3. 'Design a media-mix model for a D2C brand with 8 channels and 18 months of data.'
4. 'CAC is stable but payback period is worsening — what's happening?' (LTV shift, cohort mix, refund/return rate).
5. 'Design an A/B test for a new landing page — how many days, what sample size, what guardrails?'

SQL patterns for marketing analytics

• Multi-touch attribution join across sessions and orders.
• Cohort retention by acquisition channel.
• Funnel drop-off between step 1 → 2 → 3 with median time between steps.
• Rolling 7-day and 28-day active users.
• Revenue by first-touch channel vs last-touch channel side-by-side.

Behavioural signals interviewers look for

• Do you push back on a metric you don't believe, politely and with data?
• Can you explain a technical answer to a CMO in 90 seconds without dumbing it down?
• Do you know the difference between the analysis a stakeholder asked for and the one they actually need?