Iteration 01
M1 Activation — getting users to their first retro.
+192%
Carousel variant — winning result
99.18%
Statistical confidence — 1,974 participants
Designer specializing in complex B2B SaaS and developer-adjacent products. I prototype in production code, run end-to-end research, and ship work that moves metrics.
Two consecutive iterations on the top-of-funnel and M1 activation. Designed, tested, and implemented end-to-end — from research to production code.
2025
The signup and onboarding flow is the most critical moment in a PLG product — it's where users decide if they'll ever reach value. At Echometer, both were underperforming and untested.
What made this project different: I didn't just design the solutions — I built them. Using AI-assisted development, I prototyped directly in the production codebase, ran A/B tests with real users, and iterated between sessions before handing to engineering for production hardening.
Two consecutive iterations, each with statistically significant results.
Instead of Figma prototypes, I used AI-assisted development (Codex, ChatGPT) to build working variants directly in the codebase — HTML, Tailwind, lightweight JS. Each variant was tested with real users between sessions and iterated the same day.
Engineering then reviewed the output, corrected implementation details, and pushed to production. This collapsed the design-engineering feedback loop entirely.
