Amazon Fashion

Rich Visual Product Discovery & Personalization

Company Amazon
Role Product Lead
Scope Growth Experiments

Problem

Fashion shoppers on Amazon needed more than search results and category filters to discover products they'd love. Unlike in-store shopping where visual merchandising and curated displays drive inspiration, the online experience lacked the rich, trend-driven content that helps customers explore styles and make confident purchase decisions. The challenge was to create visually engaging, personalized product discovery experiences that could drive both customer engagement and measurable business impact at Amazon's scale.

I owned all aspects of the growth experiments end-to-end — from defining the core logic and writing SQL queries, to setting up the experimentation framework, running cohort and LTV analyses, and partnering with Marketing on paid social ads and automation to drive acquisition and retention.

Experiment

The Review Validated Programs (RVP) experience was rigorously tested through Amazon's Weblab A/B testing framework. A large-scale experiment across 13.5 million+ samples compared the treatment against control to measure impact on key business metrics. The Realm-Level Decision Recommendation confirmed statistically significant positive results across all tracked metrics, with a recommendation to launch in the US market.

Realm-Level Decision Recommendation showing A/B test results with positive impact on Composite CP, Revenue, and Paid Units
Realm-Level Decision Recommendation — A/B test results showing positive impact across Composite CP ($7.71MM), Composite Revenue ($140MM), and Paid Units ($2.79MM) with high probability of positive return

Collection

Curated visual merchandising collections were designed for the Amazon Fashion Gateway, featuring trend-driven themes that resonated with customers. The best-performing campaigns in Q4 2019 showcased how editorially curated, visually rich content could drive significant engagement — combining product imagery with compelling narratives to inspire discovery and clicks.

Best performing Gateway campaigns showing Current customer obsessions and Customer-loved styles
Best performing RVP Gateway campaigns in Q4 2019 — "Current customer obsessions" (2M impressions, 62K clicks) and "Customer-loved styles" (3M impressions, 75K clicks)

Data & Analytics

A robust data engineering pipeline was built to power the personalization and performance tracking behind the fashion discovery experience. Custom SQL-based analytics tracked product performance, customer behavior patterns, and campaign effectiveness — enabling data-driven decisions on content curation and feature optimization.

SQL data engineering code for the analytics pipeline
Data engineering pipeline — SQL-based analytics powering product performance tracking and customer behavior analysis

Customer Segmentation

The CML (Considered, Made-to-Last) Habit Forming Experience initiative identified high-value customer segments and quantified the opportunity for driving repeat purchases. Detailed segment analysis revealed which customer groups were most inclined to engage with curated fashion content, informing targeting strategies and campaign prioritization.

CML Habit Forming Experience opportunity sizing and customer segment analysis
CML Habit Forming Experience — opportunity sizing, customer segment engagement analysis, and purchase propensity by segment

Outcome & Metric Impact

The Rich Visual Product experience earned a T1 launch recommendation for the US market based on strong experimental results. The program demonstrated significant business impact with $140MM+ in expected composite revenue lift, $7.71MM in composite contribution profit impact, and measurable gains in paid units. Customer segmentation insights enabled more targeted campaigns, while the data analytics pipeline provided ongoing optimization capabilities for the fashion discovery experience.