Click an ad set to expand its ads. Meta web: only campaigns whose name contains Web are shown. Account-level financial KPIs include every Stripe purchase because this funnel has no non-Meta traffic; revenue without a captured ad ID is disclosed as Unattributed and is not guessed onto a creative. Revenue = Stripe charge-level net after Stripe processing fees, refunds and disputes. Leads = Meta Lead events attributed to each web ad; in this funnel Lead is the email submission. PostHog is used only as a fallback when Meta has no lead data for the selected period. Click→Lead = leads ÷ Meta link clicks; Lead→Paid = unique paid customer orders ÷ leads. Actual LTV = realized Stripe net ÷ unique paid users in the selected cohort. Projected LTV Day 7 / 37 / 67 = factual Stripe cohort net received so far plus expected subscription charges by that horizon, divided by the same acquired paid-user base. Current 7-day trials convert to $14.99/week: the optimistic scenario uses 45% Trial→Paid, then weekly conditional renewals of 70%, 88% and 92%; after renewal 3, 92% remains the temporary prior until our own cohorts mature. The assumption reflects the expected reduction in insufficient-funds failures at the lower charge and low voluntary churn among current successful Web payers. Legacy 7-day → $29.99/month trials already in flight retain their actual destination price and the prior monthly benchmarks 70.7%, 79.7% and 83.3%. Active Recipe Book subscriptions project their $19.99 monthly renewals with the monthly assumptions. A 3% refund reserve is deducted from every projected charge; factual Stripe refunds already reduce Actual revenue. The open dispute is deducted from Actual revenue only and is not counted as a refund. These Health & Fitness retention benchmarks come from RevenueCat 2025 and will be replaced with our web retention as cohorts mature. Other upsells stay factual and are never projected. CAC = spend ÷ unique paid users. Active trial users already paid their entry fee and are already in that denominator, so forecasting their later conversion must not add them again. There is no +30% baseline or organic uplift on this page. Unattributed payments stay outside per-creative rows instead of being assigned by guesswork. Reach / Frequency = deduplicated lifetime.