Gembah: Homepage A/B Testing and Experimentation Program
I built an experimentation program from scratch at Gembah; implementing scroll mapping, heat mapping, and hypothesis-driven A/B testing to replace opinion-based design decisions with evidence. The methodology transformed how the team approached homepage optimization across desktop and mobile.
The Problem
Gembah's homepage was underperforming on conversion, but the team lacked data to understand why. Design and copy decisions were based on internal opinions, the loudest voice in the room won. There was no testing infrastructure, no analytics baseline, and no shared language for evaluating what worked versus what didn't.
The Hypothesis
If we established a structured experimentation framework with visible data (scroll maps, heat maps, behavioral patterns), we could shift the team's decision-making culture from opinion-driven to evidence-driven, and improve conversion in the process.
The METHOD
I designed a testing program starting with baseline measurement. I implemented scroll mapping and heat mapping tools to visualize user engagement patterns, then established a hypothesis-test-learn cycle: define the assumption, design the test, run it, learn from it, iterate.
Tests covered homepage copy variations against control, mobile-specific layouts compared with desktop-adapted versions, and scroll depth analysis revealing precise drop-off points. I also conducted a technical performance audit, improving mobile site speed by 50%+. Insights fed into a broader demand generation plan connecting homepage optimization to the full funnel.
The Outcome
Building experimentation culture is harder than running A/B tests. The real challenge is getting a team to ask "what does the data say?" before "what do we think?" The scroll maps and heat maps were powerful tools for this, when you can show someone exactly where users drop off, the conversation shifts from preferences to problems.
