Weeks 1–2 = learning phase (no scaling, CVR efficiency at 70%). From week 3 onward: budget increases capped at 20% per increase, with a minimum 4-day hold between increases. Budget only scales when weekly funnel surplus exists to fund it.
Gemiini was net profitable before counting a single beta dollar. The funnel paid to acquire them.
Cold-launch CAC per paying subscriber, no funnel.
* Learning phase (weeks 1–2): Meta's algorithm needs ~50 conversion events to exit learning. CVR modeled at 70% efficiency during this period. No scaling during learning phase.
** Scaling rule: after learning phase, budget increases by 20% only when weekly surplus exists AND at least 4 days have passed since last increase. This reflects Meta's published best practices.
*** True Beta CAC = (total ad spend − non-beta funnel revenue) ÷ beta subscribers. Negative = Gemiini was net profitable before counting a single beta dollar.