$130M+ annual revenue from experimentation
Led engineering for AEO's AB testing program — including a social-proof campaign backed by a GCP data feed that drove >4% conversion lift across multiple checkout variants.
Problem
AEO had an experimentation program but the engineering org wasn't set up to push the program's most ambitious bets — high-volume tests that needed dynamic content, live data feeds, and quick iteration. The classic example: testing whether social-proof messaging and checkout "nudges" actually moved the needle.
Approach
In my final year as Engineering Program Lead for Analytics and AB Testing, I proposed and advised a plan that put a Google Cloud Platform feed at the center of the experiment. The feed dynamically delivered copy and add-to-bag data into the page, which meant variants could be iterated on without engineering rebuilds — and the data behind those variants could change in near real-time.
That same year I also led the program's migration from Google Optimize to Optimizely Web and Feature Experimentation, layering Akamai Edge Workers underneath to cut latency, improve stability, and shorten the path to production.
Outcome
- Several variants landed at roughly 4% conversion lift
- Projected cumulative impact: $130M+ in additional annual revenue for AEO
- Tooling migration shortened time-to-production for every test that followed
- The experimentation team became one of the highest-leverage engineering investments in the org
What this says about how I lead
The win wasn't just shipping experiments — it was building the engineering posture that let high-leverage experiments happen at all. Picking the right substrate (GCP feeds, edge workers, a real experimentation platform) is what turns "we test things" into "we materially move revenue."