Robust Test Designs for Dynamic Marketing, Retail, and Advertising Programs (IPM70_24Aug07)

Introduction Marketing and advertising managers began “scientific advertising” in the early 1900s with split-run newspaper tests. Large publishers printed two runs of the same newspaper. Placing different versions of their advertisement (with a unique mail-in coupon) in each run, managers could measure the number of orders from each. This offered a simple way to test new pictures, copy, prices, and other variables, simply by making one change to the ad in the second run. For the last 90 years, the majority of marketers and advertisers have continued to use the same onevariable-at-a-time testing method. The birth of design of experiments began around the same time. But as this specialized field grew, the marketing world persisted with one-variable tests, largely unaware of multivariable methods. The chasm between the science and practice of marketing testing continued. Even today, most market leaders have at best a passing awareness of scientific “multivariable testing” (as design of experiments has come to be called in marketing).