It's a safe bet that few readers of this column care deeply about the technical differences among multivariate testing methodologies. Taguchi, optimal design and discrete choice models each have strengths and weaknesses, but all are ways to quickly and efficiently identify optimal combinations of marketing treatments. The underlying technology can't be ignored totally, because it can affect practical issues such as scalability and flexibility to handle unexpected needs. But overall, users evaluate testing systems like any other software: by looking at what it would be like to use them, with little concern for what goes on under the hood. What matters is the result: better-performing promotions. Memetrics xOs (Memetrics, 415/513-5120) is a multivariate testing system based on discrete choice models. This approach measures consumer preferences by asking them to choose among versions of a complete offer, each having a different combi