How Data‑Driven Design Revamped Yixin’s Discovery Page: A Low‑Cost Redesign Case Study

Project Background

Yixin’s Discovery page, a major traffic hub, had not seen major updates for a year, leading to various usability issues. A low‑cost, data‑driven redesign was undertaken to improve user service and increase activity across social modules.

Problem Origin

Feature expansions turned the page into a promotion gateway, causing confusion such as users being redirected to the "Pengbei" tab without prior exposure, among other experience problems.

1. Understand Data

Data was collected for about a month, including server KPI data, event‑tracking (埋点) data, and related before‑and‑after page metrics. Server data provides relatively accurate usage figures, while event‑tracking offers insight into click funnels and page transitions.

2. Analyze Data

Simple comparative analyses were performed, such as comparing page click counts versus unique clickers and calculating conversion rates between preceding and succeeding pages.

3. Summarize Issues

Data analysis revealed the main problems of the old Discovery page, including low engagement for certain tabs and user confusion caused by promotional content.

Design Driven

After defining problems and product goals—boosting module usage, PV, UV, and shifting ad exposure—the team evaluated design impact, resources, and risks.

1. Design Goals and Evaluation

The page serves both as a functional entry and content showcase, allowing clear design objectives and risk assessment.

2. Design Points

Key design points were derived from research, emphasizing "minimum‑cost design"—maximizing impact with limited resources and avoiding over‑engineering.

3. Redesign Effects

Post‑launch data showed a 5‑6× surge in PV/UV for the "Shai Bei" tab due to the new content placement, though overall page PV/UV improvement was modest because core product functionality remained unchanged.

Iteration Summary

The redesign illustrates a typical cycle: identify problems, gather data, analyze, and implement solutions, while continuously reflecting on design assumptions and preparing for future higher‑cost iterations.