$12M annual revenue from personalization
As the IC owner of personalization engineering, built a data-driven framework on Adobe Target, the Customer Profile, Velocity templates, and GCP — letting business users ship personalized widgets in minutes.
Problem
Personalization was gaining momentum at AEO, and the request volume for new and varied recommendation widgets was growing fast. The early implementations treated each widget as a one-off feature injected through the experimentation tool — which caused bloat and slowed every subsequent launch.
Approach
As the individual contributor owning the engineering for personalization, I built a data-driven framework that decoupled visual treatment from data delivery:
- Adobe Target + the Customer Profile for audience targeting and decisioning
- GCP product feeds as the canonical product data source
- Velocity templates in Adobe Target Designs to render dynamic product output from those feeds
- A configurable visual layer so the design team could adjust appearance without engineering involvement
The framework's bet: separate "what to show" (decisioning + feed), "which products to show" (Velocity parsing), and "how to show it" (templating config) so any one of those layers could change without touching the others.
Outcome
- $12M annual revenue lift attributed to personalization
- Earned an Eagle's Elite award alongside Jimmy Hunkele for the Personalized Product Recommendations work
- New widgets could ship to production in minutes, not sprint cycles
- AEO landed in SailThru's Top 100 Retail Personalization Index
What this says about how I build
The lever wasn't sophisticated ML — it was a framework that let the business keep moving without engineering becoming the bottleneck. That pattern (data-driven + business-configurable + fast to ship) showed up again and again in my later work on Master Module and the experimentation platform.