Conversion in Instagram fell after a redesign. How would you investigate

Problem Statement Description

Product context: Meta is a social technology company; its products include Facebook, Instagram, WhatsApp, Messenger, Threads, Quest, creator tools, and ads. Instagram is Meta's visual social app; its products include Feed, Stories, Reels, DMs, creator tools, shopping surfaces, recommendations, and ads.

Instagram recently shipped a redesign, and the team is seeing a drop in conversion for a key user flow. You are the PM asked to investigate whether the decline is real, where it is happening, and what actions the team should consider next.

Assume the product context is Instagram within Meta’s broader social ecosystem, with particular attention to family-oriented users who may use Instagram to discover, share, message, and stay connected across generations. The redesign may have changed navigation, visual hierarchy, calls to action, trust cues, or the path users take through the experience.

This is a root-cause analysis discussion. You should frame the anomaly, clarify what “conversion” means, validate the measurement, identify affected cohorts, develop plausible hypotheses, and outline how you would use data and product investigation to narrow down the cause.

The experience should consider:

- The exact conversion event, denominator, time window, and expected baseline before and after the redesign

- Whether the drop is broad-based or isolated to specific surfaces, funnels, geographies, devices, app versions, age groups, or family-user cohorts

- Instrumentation checks, logging changes, experiment assignment issues, attribution gaps, or dashboard regressions

- Funnel step analysis to identify where users are abandoning the flow after the redesign

- Product hypotheses related to layout, discoverability, trust, performance, accessibility, or changed user intent

- External or unrelated factors such as seasonality, traffic mix, outages, ranking changes, notifications, or competitor activity

- Evidence needed to decide whether to rollback, iterate, run a follow-up experiment, or monitor further

- Prevention mechanisms such as launch guardrails, pre/post monitoring, QA checks, and cohort-specific alerting

Your goal is to show a structured investigation approach that separates measurement issues from real user behavior changes, identifies the highest-probability causes, and leads the team toward a safe, evidence-based next step without prematurely jumping to a solution.