Conversion in search and recommendations declined after a pricing or policy change. Diagnose it

Problem Statement Description

You are investigating a medium-severity conversion decline in a product’s search and recommendation surfaces after a recent pricing or policy change. The affected users are analysts who rely on search results and recommended items to complete work tasks efficiently, such as finding relevant datasets, reports, products, templates, vendors, or content depending on the product context.

The decline is visible in downstream conversion from search and recommendations, but the root cause is not yet known. It may be related to user behavior changes, ranking or eligibility shifts, price visibility, policy restrictions, inventory/content availability, tracking issues, experiment exposure, or changes in the mix of users and queries.

Your task is to diagnose the anomaly in a structured way. Focus on how you would frame the problem, validate the metric movement, segment the impact, form hypotheses, identify evidence needed, and decide what immediate mitigations or follow-up actions are appropriate.

The experience should consider:

- How conversion is defined for search and recommendations, including numerator, denominator, funnel step, attribution window, and whether “task success” differs by analyst use case.

- Whether the decline is real or caused by instrumentation, logging, attribution, data pipeline, experiment assignment, or dashboard definition changes.

- Segmentation by surface, query type, recommendation module, user cohort, geography, device, account type, pricing tier, policy eligibility, and new versus returning users.

- The timeline of the pricing or policy change, rollout scope, communications, enforcement rules, and any concurrent launches or ranking changes.

- Funnel diagnostics across impressions, clicks, engagement, add-to-cart/save/apply/contact actions, checkout or completion, and abandonment points.

- Hypotheses around user trust, perceived value, affordability, result relevance, restricted availability, reduced recommendation diversity, or unexpected friction.

- Evidence needed from analytics, qualitative feedback, support tickets, search logs, recommendation traces, A/B tests, and operational metrics.

- Immediate mitigation options, monitoring needs, stakeholder communication, and prevention mechanisms for future pricing or policy changes.

The goal is to demonstrate a clear RCA approach that separates correlation from causation, protects user task success, and gives product, analytics, engineering, and business teams a practical path to understand and respond to the conversion drop.