A book that first got me interested in pursuing Data Science was Claude C. Hopkins’ Scientific Advertising. Hopkins pioneered the use of statistical testing and test campaigns in the field of advertising. While Hopkins focused on print advertising and supermarket mailers, many of his followers like David Ogilvy would expand his concepts to television and later to the internet. Tracking media consumption was essential to how publications and channels would value their advertisement. Just as a restaurant in Milwaukee would not want to advertise in Tallahassee, knowing who a channel would reach was just as important as learning the audience size.
Target had an interesting case of targeting a customer in 2012. Almost half of our daily decisions are habitual decisions rather than conscious decisions, so reaching soon-to-be parents early-on is one of the holy grails for advertisers, as this group most often totally commits to a brand. Through analyzing changes in buying habits, the statistician Andrew Pole was able to design a model that helped Target predict soon-to-be parents before their child was due. The model was briefly in the popular conscious when its accuracy led to an angry father in Minneapolis berating the manager his local Target for mailing his high school-aged daughter advertisement, only for him to find out that his daughter was in fact pregnant. Because the practice made some customers queasy, Target revised their coupon distribution to appear random, finding that as long as the pregnant woman felt she wasn’t being spied on, she’d buy the desired brand.
While supermarket stores like Target use data gleaned from a customer’s participation in loyalty programs and credit cards in conjunction with purchased personal data like ethnicity, salary, birth records, and voter affiliation to predict and encourage consumer habits, I was interested to learn how tracking media consumption began.
Interestingly, modern consumption tracking was begun by the same company that helped produce a consumer price index for supermarkets, Nielsen.Like many, I always assumed this form of data collection was performed exclusively through phone calls. While this was the primary method employed by their competitor, Hooper Ratings, Nielsen employed a device called an Audimeter that attached itself to select radio and television sets andwould report at what times the set was on and to what stations the set was tuned. Prior to the use of cable television boxes to report viewer habits, the Audimeter was used in a sample of American homes and made its recordings of viewer habits on a 16mm film. When the film was filled, the film cartridge would eject and provide the consumer with a quarter so that they could mail the cartridge to Nielsen for reporting.
Not only did Nielsen ratings allow networks to better price advertisements and shape creative content, but from it, it would let them know if an advertisement actually reached the right group. Perhaps because of their large investment in Univac computers, they were able to extrapolate valuable insights from this large body of data, not only for their partnered brands but for television and radio writers as well.