Experiment design to decide on launching ACH bank payments for Netflix US members

Netflix wants to let members in the United States pay by ACH bank transfer, and you are helping with the launch. Design the test you would run to decide whether ACH should become generally available. ```hint Find what could make ACH a bad deal ACH is cheaper to process than cards, but it behaves differently after the charge. Work out what the test must observe to see both the savings and the costs that arrive later. ``` ### Clarifying Questions - Is Netflix's main reason lower processing costs, reaching people without a usable card, fewer failed renewals, or a mix? - Would ACH be offered at sign-up, to existing members in payment settings, or both, and would members get an incentive to switch? - How do members link their bank accounts (instant verification by bank login or micro-deposits), and who bears the losses on returned payments? - What does generally available mean here: offered to every US member, or available but not promoted? ### What a Strong Answer Covers - Why Netflix wants ACH, translated into the decision metric for general availability - A randomized design that compares members offered ACH with members not offered it, rather than ACH payers with card payers - ACH-specific outcomes: bank-linking drop-off, returned payments, unauthorized-debit claims and their delay - Sample size, a test length that covers several billing cycles and late returns, and a staged ramp - Launch criteria and guardrails agreed in advance, plus operational readiness ### Follow-up Questions - The test shows lower processing costs but higher churn among members who pay by ACH. How do you tell whether ACH caused the churn? - Should Netflix share the savings with members through a discount for paying by ACH? How would you test that? - After launch, which signals would make you pull ACH back?