ZG

Course Vaults

Building Course Vaults from idea to global golf platform

Role

Founder

Product Designer

Frontend & Backend Engineer

Platform

Web & Mobile (B2C)

Skills

Product Architecture, Design, Rankings, Subscriptions, Data Operations

How I designed, built, launched, and monetized a web and mobile product covering nearly 40,000 golf courses, with 7,000+ profiles, 226,000+ ratings, and a growing subscription business.

Courses
40,000
Profiles
7,000+
Ratings
226,000+

What is Course Vaults

Course Vaults is a web and mobile product for golfers to track and rate the courses they’ve played, a shared, player-driven rating system of courses around the world.

Instead of panels, critics, or curated rankings, it’s powered by everyday golfers. Each rating adds to a collective view of the course. Letterboxd for golf, designed and shipped end to end.

The product

I founded Course Vaults to replace critic panels and curated lists with a graph built by golfers. People search the map, save courses, rate what they’ve played, keep a diary, and see how their circle scored the same places.

The surface area is the whole product: discovery, rankings, mapping, community, subscriptions, and the data operations underneath. I designed it, implemented it on web and mobile, and still ship it as a team of one, often with AI-assisted development.

Course Vaults profile and course detail screens

Activation

Early on, 28% of users never rated a course. The rating flow asked for too many decisions before anyone got value. I cut it to a one-page experience and asked new users to rate familiar, nearby courses first.

Empty profiles fell from 28% to 8%, a 71% reduction. People could build a real diary in the first session.

Reducing the cognitive cost of contribution

Monetization

Onboarding worked. Usage spiked, then dropped. The ask to pay was landing after intent had cooled. I gated the free plan at 20 saved courses so the paywall shows up while someone is still adding courses they care about.

Free-to-paid conversion moved from 4.5% to 5.85%. That is a 30% lift. Early, and the right threshold is still an open test.

Conversion is a timing problem

What I own

Architecture, rankings, mapping, community mechanics, subscriptions, and the data layer that keeps 40,000 courses current, including AI-generated descriptions so coverage can grow without writing every page by hand.

2025-Now