Investigate Falling Brand-Ad Spend
Company: Meta
Role: Data Scientist
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Technical Screen
On a video ads platform, there are two ad products:
- Direct Ads: optimized for immediate response.
- Brand Ads: when a user clicks the video ad, they are taken to the advertiser's website.
A large advertiser reports that spend on Brand Ads has fallen materially over the last few weeks. How would you:
1. Validate that the decline is real rather than a dashboard, tracking, FX, or pipeline issue?
2. Quantify the size of the drop relative to an appropriate baseline?
3. Diagnose the root cause?
In your answer, explain what data you would pull, what baselines or control groups you would use, how you would segment the analysis, and which metrics you would inspect. Consider multiple plausible explanations such as seasonality, budget changes, auction competition, targeting changes, inventory shifts, creative fatigue, landing-page problems, attribution or tracking bugs, and cannibalization between Brand Ads and Direct Ads.
Quick Answer: This question evaluates competency in data analysis, anomaly detection, causal inference, and diagnostic reasoning related to advertising performance, within the Analytics & Experimentation domain.