Company Context

Meta operates one of the largest consumer recommendation surfaces in the world. Facebook News Feed serves billions of ranked impressions daily and is a core driver of engagement, ad supply, and downstream actions across the Meta ecosystem, including reactions, comments, shares, profile visits, and session depth.

Problem

Last Tuesday, Facebook News Feed click-through rate (CTR) dropped 8% day-over-day. Leadership wants a product-focused investigation, not just a dashboard readout: determine whether this is a measurement issue, a ranking or inventory issue, a user-behavior shift, or an intentional trade-off caused by another product objective. Assume you have access to Meta’s standard product analytics stack, experiment logs, feed ranking launch history, integrity interventions, and segmented metrics across surfaces such as Feed, Stories, Reels distribution modules, and Facebook Groups recommendations.

Your task is to frame the investigation as a product sense problem: identify which user experiences may have degraded, where in the aarrr funnel the issue matters most, and what decisions you would make if the CTR decline came with mixed movement in other metrics like time spent, meaningful social interactions, ig save on cross-posted content, or long-click quality.

Deliverables

1. Define how you would clarify the metric and scope the problem before proposing causes.
2. Segment the drop across users, content types, ranking surfaces, and geographies to identify the most likely user need that broke.
3. Prioritize the top hypotheses you would investigate, including instrumentation issues, Sample Ratio Mismatch (SRM), launch side effects, and product trade-offs.
4. Explain what actions you would recommend under at least two different findings (for example: bug, bad launch, or healthy trade-off).
5. Specify the success criteria and guardrails you would use for any mitigation or rollback decision.

Constraints

• You have 24 hours to provide an initial recommendation to Feed leadership.
• Multiple launches shipped in the prior 7 days, including ranking tweaks and integrity changes.
• You cannot optimize CTR in isolation if quality, safety, or long-term retention would be harmed.
• Any follow-up experiment should use Meta-standard experimentation practices, including CUPED where appropriate and checks for novelty effect and SRM.