The company was fictional, and so were its three products: Culp, Sunbars and Digits. Its stated objective was “to maximize profits,” and the case added: “you agree with this.”
Culp had ten times the market share of its leading competitor in a market growing two percent a year. Sunbars had a relative share of 1.5 in a market growing 20 percent a year. Digits had about a fifth of its leader’s share in a market that was not growing at all, and barely adequate profits.
Two proposals were on the table, and only one could be funded. An inventor had found a cheaper way to make Digits: invest $1.5 million, save about $500,000 a year after tax for ten years. The Sunbars division wanted an advertising campaign: invest $1.8 million, earn $400,000 the first year, $300,000 the second and $100,000 a year for eight years after that. The case said the forecasts could be trusted and that neither choice had side effects.
J. Scott Armstrong of the Wharton School and Roderick Brodie of the University of Auckland built the case so the answer would be obvious. In rough terms, counted at present value with a 12 percent cost of capital, the Digits money would double over ten years. About half the Sunbars money would be lost. They also built it so that the Boston Consulting Group’s growth-share matrix would point the other way. Digits was a Dog. Sunbars was a Star.
From 1985 to 1990, six faculty members ran the exercise on 27 occasions in six countries, with 1,015 people from management programs: undergraduates, MBA students with work experience, and middle managers. Each person worked alone, and versions of the materials were assigned at random. Some got only the case. Some were told they had recently attended a course on the BCG matrix and were handed a summary of it from a widely read marketing textbook. Others got a short note on net present value.
The control group did worse than the authors expected: 44.7 percent chose Sunbars. With the NPV note, 37.1 percent did. With the BCG summary, 63.5 percent chose the Star.
Of the 296 people given the matrix, 173 used it explicitly. Nobody assigned that part; people decided for themselves which method to use. Across all groups, 86.8 percent of the people whose analysis used the BCG matrix chose Sunbars, against 15.3 percent of those who worked from profit figures. “Only 13% of those who used the BCG matrix in their analysis invested in the more profitable project,” the authors wrote. That split was sorted by method after the fact, so unlike the comparison between handouts, it is not a randomized result.
They checked whether people were simply obeying the handout. In later sessions a control group got both the BCG and the profit-based notes, and 60 percent of them chose the Star. When the authors swapped which product had the advertising plan and which had the invention, the result held.
The authors listed the limits. It was a laboratory task, deliberately built so the matrix and the profit figures conflicted, and it says nothing about how often such conflicts come up in real companies. One of them, they noted, had been an advocate of the matrix when the study began. The paper was published in 1994 with a critique from Robin Wensley in the same issue. As the authors summarized it, Wensley argued that the study was flawed. In their reply, Armstrong and Brodie agreed it was flawed, “but not to the extent that Wensley states the case.”
Most people spent more than 20 minutes on the exercise, the administrators reported, and in the debriefing sessions afterward, they defended their decisions vigorously.