Prop GPT

Eyal and Yali launched Prop GPT mid-NFL season and immediately leaned on influencer marketing to drive downloads. They spent a few thousand dollars on influencer sponsorships, which generated roughly 20 downloads per day. Of those, 5–10 per day converted to free trials, giving them a strong download-to-trial rate early on. This gave them their first paying users and got them to ~$1,000–$2,000 MRR relatively quickly. However, despite healthy download and trial numbers, conversion from trial to paid was extremely low (only ~13%). They diagnosed this through analytics tools (PostHog) and user behavior data: on high-traffic moments like the Super Bowl, they saw spikes in trial starts but almost zero retention after trial. This was the signal that their first-user playbook was working, but the product itself was failing to deliver on its promise.

Eyal had prior experience working with successful app founders and had already observed the influencer marketing playbook in action before building Prop GPT. This gave them a ready-made distribution channel from day one — most first-time founders spend months figuring out what they already knew.

influencer_marketing

The original version of the app required users to manually input their bets and check them against the model — users wanted to simply be given the best picks directly. This UX mismatch caused trial-to-paid conversion to stagnate at ~13%, capping MRR at $1,000–$2,000 despite healthy download volumes. All marketing was shut down during the 4-month rebuild because pouring spend into a broken product was pointless.