Convince Product Manager to Launch 'Show Similar Products' Button
Quick Overview
Meta data scientist product analytics prompt on using proxy metrics and experiment design to test an Instagram "Show similar products" button, including guardrails, randomization, power, and rollout criteria.
Convince Product Manager to Launch 'Show Similar Products' Button
Company: Meta
Role: Data Scientist
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Technical Screen
##### Scenario
Instagram is considering adding a 'Show similar products' button to boost user engagement, but the feature has not launched yet.
##### Question
How would you convince the Product Manager that the feature is necessary before any launch data exists? Using only the existing interaction data, what proxy metric(s) would you choose to estimate the feature’s potential impact on engagement? How would you design an experiment to decide whether to launch the button, including randomization unit, control versus treatment, and guardrail metrics? What criteria would signal it is safe to roll the feature out broadly?
##### Hints
Link metrics to engagement, propose historical baselines, cluster randomization to reduce network effects, set significance level and guardrails for health metrics.
Quick Answer: Meta data scientist product analytics prompt on using proxy metrics and experiment design to test an Instagram "Show similar products" button, including guardrails, randomization, power, and rollout criteria.
Convince a PM to Test a "Show Similar Products" Button
Instagram is considering adding a "Show similar products" button on product-tagged content to boost shopping-related engagement. The feature has not launched yet.
Use existing interaction data to decide whether the feature is worth testing and to design an experiment for the launch decision.
Constraints & Assumptions
No direct feature data exists yet.
Use proxy metrics carefully and explain why they map to similar-product intent.
Separate evidence for testing from evidence for full launch.
Include guardrails for overall app health and shopping quality.
Clarifying Questions to Ask Guidance
Where would the button appear: feed, stories, reels, product detail pages, or shops?
What is the target outcome: engagement, product discovery, clicks to merchant, purchases, or revenue?
Which users and product-tagged surfaces are eligible?
Are there known concerns about feed clutter, latency, or creator/merchant fairness?
What a Strong Answer Covers Guidance
Proxy metrics from existing logs, such as product-tag taps, related-search behavior, shop/profile pivots, saves, shares, add-to-cart, product detail views, and repeated category browsing.
Historical baselines and simple projections, including lower and upper bounds based on current intent signals.
A test design with eligibility, treatment/control definition, user or cluster randomization, exposure logging, and ramping.
Primary metrics tied to incremental shopping engagement or conversion.
Guardrails such as retention, session quality, feed engagement, latency, hide/report rate, merchant quality, and cannibalization.
Power, duration, significance level, and decision thresholds for rollout.
Criteria for global rollout, targeted rollout, iteration, or rollback.
Follow-up Questions Guidance
Which proxy metric would you trust most before launch, and why?
How would you handle network effects or creator-side spillovers?
What would you do if clicks rise but purchases do not?
How would you decide between global launch and segment-targeted launch?