A two-stage acquisition funnel for high-consideration, high-average-order-value DTC products, used by Flooret (average full flooring order ≈1,000 sq ft, comparable in dollar value to a car).
Stage 1 — sample acquisition: paid acquisition drives customers to buy an inexpensive cut sample of the product, priced as a low-friction, low-risk entry point rather than to be profitable on its own.
Stage 2 — nurture to full purchase: sample buyers are nurtured with design and education content (email/SMS, content marketing) toward the full-size purchase, which is where the real order value and margin sit.
Economics: success is judged on aggregate customer lifetime value across the whole funnel, not on sample-order profitability — the brand can lose money on most individual sample orders as long as enough of them convert into full purchases to make the blended funnel profitable. This mirrors the "spend below what a customer is worth over their full relationship" logic behind LACoS (Lifetime ACOS) and Lifetime TACOS, applied outside the Amazon-ads context.
Relevant when porting this model to a marketplace: Amazon does not support the same free-form quantity selection used on a brand's own site, which changes how the sample/full-size split has to be merchandised (see Preemptive SKU Quantity Configuration for Variable-Quantity Products (Amazon Parent-Child ASINs)).
The funnel is tracked with two core KPIs: (1) cost to acquire a sample buyer (CAC-to-sample) and (2) the conversion rate from sample buyer to full-size purchaser. Neither KPI alone is sufficient — a cheaply-acquired sample buyer is worthless without a high enough conversion rate, and vice versa. Because the sample itself is a loss-leader, paid acquisition spend on it is evaluated against blended customer LTV across the whole funnel rather than the sample's own standalone margin: the entry offer can lose money on most units as long as aggregate LTV from converters justifies the total acquisition cost. See also Customer Lifetime Value (LTV) as the Primary Amazon Margin Lever.
Alvaro Lopez (Flooret) treats the physical sample itself as an intentional loss-leader, structurally similar to a freemium software model: most samples never convert, and that's expected. Success is judged on blended customer LTV across the whole population of sample requesters, not on the margin or conversion rate of any individual sample.
Apply: don't evaluate a sample program by per-sample conversion rate or cost alone — model it like a freemium funnel and judge it on blended LTV across all requesters, non-converting samples included.