John Wanamaker, the department store pioneer, famously said in 1886, "Half the money I spend on advertising is wasted; the trouble is I don't know which half." In the early decades of the 20th century, launching an industry and anticipating issues in knowledge management, business and advertising executives set out to tell Wanamaker's successors not only which half of their money worked, but why.
Modern market research was born shortly before World War I at the Curtis Publishing Co., a magazine publisher. Cyrus Curtis was one of the leaders in the shift that began in the 1890s from reader-supported to advertiser-supported publications. As advertising sales became increasingly important, Curtis hired a one-time Wisconsin high school principal, Charles Coolidge Parlin, to set up and head the company's new consumer research division.
Parlin's charge was to do "commercial research"- -to develop data to help the company understand the U.S. consumer market. The goal was to help Curtis's ad salespeople sell ads more effectively in three key magazines: Country Gentleman, Ladies' Home Journal, and Saturday Evening Post.
For Saturday Evening Post, for example, in 1912 Parlin's staffers visited every medium-size and large city in the U.S. In each they buttonholed merchants and asked questions about their sales and their customers. The resulting survey for the first time reported per capita expenditures on dry goods and women's clothing; it also provided hard numbers on costs, turnover and profit in the retail trade. (In this report Parlin wrote that "The Consumer is King" and also declared that woman is a shopper and "a comparer of values.") Parlin's team later surveyed automobiles (1914), food products and household supplies (1914-1917), linoleum (1915), canned soup (1917) and canned beans (1919).
These studies increased advertising revenues for Curtis magazines; they also established commercial fact-finding as a profession and a legitimate business undertaking. In 1916, for example, the United state Rubber Co. established a research department under Paul Nystrom, who taught at the University of Minnesota and later at Columbia University. One year later meat-packer Swift and Co. commissioned L.D.H. Weld, a Yale economist, to develop a commercial research department.
Academe did more than simply contribute experts, and the already thriving advertising industry played a role as well. The Harvard Graduate School of Business established a Bureau of Business Research in 1911, and four years later the National Association of Teachers of Advertising (ultimately part of the American Marketing Association) was formed. Psychologist Walter Dill Scott was NATA's first president, and the second was Paul Cherington, one of the first teachers of marketing at Harvard's Graduate School of Business and also a head of the J. Walter Thompson agency's research department. JWT also was for a while home to John B. Watson, the Harvard psychologist and father of behavioral research.
Systematically polling the users of products themselves to determine consumer attitudes was introduced in the form of a series of nationwide surveys conducted by psychologists from the Psychological Corp., a company formed in 1921. Its original mission was to develop psychological tests and materials for practicing psychologists. In the process of this work, though, the corporation began to ask consumers' opinions not only about their lives, work and public issues, but also about cars, coffee, and cigarettes.
Nothing symbolized the ascendancy of scientific data sampling more dramatically than the widely reported debacle of the Literary Digest 1936 presidential poll. For 20 years the magazine had successfully forecast election results. In 1936 the Digest sent out more than 10 million ballots to people who had subscribed to a telephone service or owned a car; more than two million returns were tabulated. The result was a fiasco. The survey predicted that the Republican candidate, Alfred M. Landon, would beat Democrat Franklin D. Roosevelt, and it overestimated Landon's vote percentage by some 20 percent. Roosevelt was re-elected in a landslide.
The poor showing by the 1936 Digest poll stemmed from its failure to guard against a biased sample of replies. At that time, telephone subscribers and automobile owners tended to be more affluent and to vote Republican, so less-affluent and less-educated voters were badly underrepresented. In that same election, three polling newcomers using the new scientific methods accurately forecast the result. These were George Gallup's American Institute of Public Opinion, the Fortune Survey, conducted by Cherington and Elmo Roper, and the Crossley Poll, under the direction of Archibald Crossley. All achieved their results by means of personal interviews with small but selected samples of voters- -a model for the scientific data collection and opinion sampling that has characterized the last six decades of market research.