Diagnose a sudden drop in learning completion for live event discovery

Problem Statement Description

You are investigating an unexpected drop in learning completion among field operators who use a live event discovery product. These operators rely on the product to find relevant live sessions, trainings, briefings, or operational events and complete required learning as part of their day-to-day work.

The issue is framed as a sudden metric anomaly, not a request to jump directly into fixes. Your task is to diagnose what changed, where the drop is concentrated, and whether the problem is caused by user behavior, product experience, content availability, measurement issues, operational changes, or external factors.

Assume the product operates at scale across different regions, devices, shifts, and operator roles. Field operators may have constrained time, variable connectivity, mobile-first workflows, and dependencies on schedules, reminders, eligibility rules, and event availability.

The experience should consider:

- How “learning completion” is defined, including the denominator, completion criteria, time window, and whether partial progress is counted.

- Whether the anomaly is real or caused by tracking, logging, data pipeline, event taxonomy, or dashboard changes.

- Segmentation by operator role, geography, device type, app version, language, shift, network quality, and learning/event category.

- Funnel steps from event discovery to registration, attendance, content consumption, assessment, and completion recording.

- Recent changes to ranking, search, recommendations, notifications, eligibility rules, event inventory, scheduling, or required-learning policies.

- Evidence needed to distinguish product friction from content supply issues, operational constraints, or seasonality.

- Immediate mitigations to reduce user impact while the root cause is being validated.

- Longer-term prevention through monitoring, alerting, data quality checks, and ownership of critical completion flows.

Your goal is to structure a clear RCA approach: confirm the metric drop, localize the affected population and funnel stage, generate and prioritize hypotheses, identify the evidence needed to validate them, and outline how you would communicate findings and next steps before recommending fixes.