Caleb Kim User research and A-B testing through Conch AI's early growth — 2M+ users reached.
Conch AI is an AI writing tool that reached 2M+ users during the early growth I contributed to. As UX Researcher (2023–2024) I ran the loop between a fast-growing Discord community and the product team — mining feature boards and idea contests, A-B testing live in Statsig, and turning it all into weekly reports the design and dev teams ran on, across a 150K+ active user base.
Conch rode the first AI-writing wave — the community Discord went from 500 to 2,000 members in about ten days of March 2023, and the product went on to reach 2M+ users. At that speed there's no time for research theater; the job was keeping product decisions connected to what users actually wanted.
I ran the feedback loop between the community and the team: mined the feature boards and idea-contest results, DM'd power users, set up community mods to triage issues, and brought the asks with real merit to the table — including input that shaped visual design. Everything worth acting on landed in weekly Notion reports the design and dev teams worked from, and design questions were settled with live Statsig A-B tests — color schemes, button placements, whole features.
The loudest ask is not always the right one — the real work was sorting merit from noise, and a voting community does half of that sorting for you if you build it the pipeline. In a ~10–15 person team with no fixed lane, the weekly report turned out to be the product: the one place where two thousand voices became a decision.
“Conch's growth had many drivers — product, marketing, timing. My lane was narrower: finding what users actually wanted and making the case for it.”