The problem
On campus, startup ideas often die at the team-formation stage. The person with the idea cannot find the person with the skills, and the clubs or Discord servers where they might meet are organized around everything except that matching problem. Lokin exists to make finding a startup team feel as direct as finding an internship.
What we built
A matching platform where student founders post projects with the roles they need, and students looking for real startup experience apply to join them. I co-founded it and led product across research, roadmap, and much of the build, backed by the Gener8tor-backed CS Nest Accelerator. We grew to 500+ signups in the first two months through campus outreach, club partnerships, and a startup conference.
The funnel story
Early signups told a misleading story. Growth looked fine while activation was broken. Only 5% of people who signed up completed a profile, and an unmatched profile is worth nothing on a matching platform. Funnel analysis pointed at onboarding, a six-step profile flow that asked for everything upfront. We rebuilt it as three steps, and completion moved from 5% to 50% of signups.
The chatbot that was secretly a research tool
I built a RAG-powered onboarding chatbot to answer questions during signup, and 95% of new users touched it. The useful part was not just support. It became a passive feedback channel. Reading what people actually asked surfaced patterns no survey caught, like users wanting to search by role rather than by project. That changed how we thought about matching.
What I learned
Growth metrics flatter you. Activation metrics tell the truth. Every instinct early on was to add features and widen the funnel, and the right move was the opposite: fix the three screens where users actually were.
With two founders there is no backlog to hide behind. Every week becomes a forced ranking of what matters. That constraint taught me more about product prioritization than any framework.