Frank Smith, an independent sports creator with 4 million followers, built GeoSports — a daily sports geography web game — in roughly 8 hours over a weekend using Claude; three months later it generates about $40,000 per month with 700,000 monthly active users and a 53% organic share rate.
What GeoSports is and how it works
GeoSports (part of the Geo Games suite with Geo History and Geo Footy) is a browser-based game where players get five clues per day and tap the correct location on a globe.
No tutorial needed: the mechanic is intuitive — tap where the event happened, get scored by proximity (green dot for close, yellow for far), repeat for five questions, see results.
The game lives entirely on the web, so players can open, play, and share without leaving platforms like X; near-zero friction is a core growth lever.
Revenue breakdown
~$40K/month total: ~$30–35K from programmatic ads (roughly $1,200/day, still ramping), ~$5K from ~1,800 Pro subscribers via Stripe.
Ads are the dominant stream; subscriptions are a smaller, newer layer.
Founder background and prior attempts
Frank has been a full-time sports content creator since 2020 (started during COVID), known for high-quality short-form commentary videos.
He has always loved games: launched Five Card Draw (~110K players) and several other niche games before GeoSports.
GeoSports was not the planned launch; a different game was in the works, but a weekend experiment with a sports twist on MapTap took off unexpectedly.
Building the app in 8 hours (non-technical, AI-assisted)
Frank does not know how to code. He started with a detailed spec, asked Claude to research MapTap, then iterated by pasting errors and screenshots back to the LLM.
Worked ~8 hours on Sunday, cleared Monday to keep going, and had a playable version to share within days.
Tech stack: Claude Max (20x), Vercel Pro (hosting + custom domain), MapLibre (globe), Upstash Redis (leaderboards/stats), Google Sheets (question database), Stripe (Pro subscriptions).
Lean, serverless, and entirely managed through AI guidance — screenshots and constant “is this right?” checks were the workflow.
Growth: zero to $40K/month in three months
Stealth launch to friends/family → noticed organic sharing → posted a reply to his own unrelated X tweet (“by the way I made this game”) → reply went viral: 40K players day one, 150K day two.
Unprepared for retention at 5 days old; scrambled to stabilize.
Core growth engine: word-of-mouth, not audience. Only ~2% of daily traffic (~2K clicks) comes from Frank’s own daily X post + Instagram story; 98% is players sharing in group chats, unaware of the founder.
Hosting role cheat code: created a “host” slot for sports media personalities and athletes. They write the five daily questions, Frank announces them, hosts retweet → new audiences daily. Feels like a cheat code; also builds Frank’s network.
Habitual daily play + high shareability (53% of players share their score) drives sustained viral loops.
Advice for builders in 2026
Build for web first. Zero install friction lets people play, share, and return instantly from social feeds; native apps add drop-off.
Make the product inherently sharable. GeoSports succeeds because the result screen begs to be posted; 99% of traffic is organic word-of-mouth, not creator promotion.
Don’t wait — block 8 uninterrupted hours and start. Getting started is the hardest part; when stuck, ask the LLM, screenshot errors, keep moving.
Use screenshots and relentless verification. Frank treated the LLM like a pair programmer: “here’s my screen, what am I looking at?” → “is this config right?” → iterate.
Final advice: stay in the game
Frank’s YouTube journey: started 2007, first real traction 2020 (~13 years). Game projects started 2020, breakout 2026 (6 years later).
Patience and persistence compound; none of this happens if you quit during the long flat stretches.
