Another factor that appears to be ignored in this report on the study is the perceived multiplier of the transaction delta from repeat business. If I'm going to save 65 cents on every book Amazon sells me in exchange for surrendering my (same) email address every time, that's very different from a one-time only discount or a unique purchase from a vendor I'm unlikely to revisit. As it happens, the full report does mention two models, one with and one without multiple transactions, but without reading all 76 pages, it's not clear how the 65 cent figure relates to those more complicated situations.