How to evaluate Shop ad upranking
Company: Meta
Role: Data Scientist
Category: Analytics & Experimentation
Difficulty: hard
Interview Round: Technical Screen
Meta is considering ranking Shop ads higher on its consumer surfaces so that users can complete purchases inside Meta Shops instead of being sent to an advertiser website. The product hypothesis is that in-app checkout may reduce friction and help smaller advertisers that do not have strong websites, but it could hurt larger advertisers whose goals may include website traffic, app installs, or offline store visits.
How would you evaluate whether upranking Shop ads is a good idea?
Address the following:
- Define primary success metrics across user value, advertiser value, and platform value.
- Propose guardrail metrics that would detect harm to user experience or advertiser outcomes.
- Design an experiment, including randomization unit, treatment definition, test duration, and how you would handle seasonality and interference.
- Explain how you would distinguish true incremental value from simple channel substitution, such as Shop purchases replacing website purchases.
- Discuss how you would analyze heterogeneous effects for small versus large advertisers and for different advertiser objectives.
- State the decision rule you would use to recommend a full launch, a limited rollout, or no launch.
Quick Answer: This question evaluates a data scientist's competency in causal experimentation, metric design, uplift and channel-substitution analysis, heterogeneous treatment-effect estimation, and trade-off reasoning across user, advertiser, and platform value.