When evaluating the net impact of a series of percentage changes, we predict that consumers may employ a "whole number" computational strategy that yields a systematic error in their calculation. We report on three studies conducted to examine this issue. In the first study we identify the computational error and demonstrate its consequences. In a second study, we identify several theoretically driven boundary conditions for the observed phenomenon. Finally we demonstrate in a real-world retail setting that, consistent with our premise, sequential percentage discounts generate more purchasers, sales, revenue, and profit than the economically equivalent single percentage discount.