How investigate a brand-ad spend drop?
Company: Meta
Role: Data Scientist
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Technical Screen
Meta has a video ads product with two ad types:
- **Direct ads**: optimized for in-platform actions
- **Brand ads**: users click the video ad and are sent to the advertiser's external website
One advertiser reports that its **Brand Ads spend** has dropped materially over the past two weeks. As a data scientist, explain how you would:
1. Validate whether the spend decline is real rather than a logging, reporting, or seasonality artifact.
2. Determine whether the issue is isolated to this advertiser or reflects a broader product or system change.
3. Identify the most likely root cause.
4. Recommend follow-up analyses or experiments.
In your answer, define the key metrics you would examine, show how you would decompose spend, discuss plausible confounders, and explain how you would avoid misleading conclusions caused by placement mix, geography mix, or attribution differences.
Quick Answer: This question evaluates a data scientist's competency in diagnosing ad-spend anomalies, including data validation, attribution and reporting integrity, segmentation and placement effects, and causal inference within advertising analytics.