You are in charge of investments at a company called Conglomerate, Inc. Its objective is to maximize profits, and you can fund one of two proposals.
The first is for Digits, a small product in a market that isn’t growing. An inventor has found a cheaper way to make it: put in $1.5 million and save about $500,000 a year after tax for ten years. The second is an advertising campaign for Sunbars, a market leader in a market growing 20 percent a year: put in $1.8 million and earn $400,000, then $300,000, then $100,000 a year for eight years.
In the authors’ rough terms, counted at present value with the 12 percent cost of capital the case’s finance note used, the Digits money doubles over ten years and about half the Sunbars money is lost. J. Scott Armstrong and Roderick Brodie wrote the case that way on purpose, and between 1985 and 1990 six faculty members gave it to 1,015 people in management programs in six countries.
Some of them had been handed a summary of the Boston Consulting Group’s growth-share matrix first. On that matrix, Digits was a Dog. Sunbars was a Star.
Of the people given the summary, 63.5 percent chose the Star.
Following the cash
Growth and share show where cash comes from and where it goes. A label isn’t a verdict.
Bruce Henderson, who founded BCG, laid out the idea in a 1970 essay called “The Product Portfolio.” High market share, he argued, goes with high margins, and growth eats cash because a growing business needs new assets to hold its place. Put those together and every product lands in one of four boxes. A slow-growing leader throws off more cash than it can usefully reinvest: a “cash cow.” A fast-growing leader, a “star,” will become a cow when growth slows if it holds its lead. A small player in a fast market, a “question mark,” almost always needs far more cash than it makes. And a small player in a slow market was, in Henderson’s original word, a “pet”: “essentially worthless, except in liquidation.” In much of what was written later, including the studies below, it became a Dog. “Pets are not necessary,” he wrote.
For a company running dozens of divisions, that was a useful way to see the whole. BCG’s own 2014 look back says the matrix gave conglomerates a logic for moving cash out of mature units, which often kept and reinvested their own cash, into units that could grow.
But a box is a summary of two numbers, and a decision needs more than two. Armstrong and Brodie suspect the labels work like mental accounts. A Cash Cow sounds like money to take out, a Star like money to put in, a Dog like money not to spend. In their experiment, 86.8 percent of the people who reasoned with the matrix picked the Star, against 15.3 percent of those who worked from the profit figures.
A business called a Dog sounds like one to get rid of. The name can make that sound settled before anyone has looked at what the business earns.
When a Dog pays
The matrix’s own premise has been tested too. Donald Hambrick and Ian MacMillan, working with data on industrial-product businesses in the PIMS database, found in 1982 that the four boxes did differ in how they generate and use cash, as BCG said. The dogs were the exception. In their sample the average Dog had a positive cash flow, larger than the cash needs of the average question mark (which they called a Wildcat), and some Dogs did very well.
In 2014 BCG’s own researchers went further. Looking at US public companies, they found businesses moving between boxes faster than before and concluded that “market share is no longer a strong predictor of performance,” while arguing that the matrix, updated, still has value.
The experiment has limits worth stating. It was a classroom exercise built so the matrix and the profit figures would disagree, and the authors say it can’t tell how often that happens in real companies. The handouts were assigned at random, so 63.5 against 44.7 percent is a comparison between groups. Which method people actually used was their own choice, so the 86.8 against 15.3 percent split, sorted afterward by method, is not a randomized comparison. A critic, Robin Wensley, commented in the same journal issue. As the authors summarize him, he argued the study was flawed, and they agreed, though not to the extent he claimed. Even the control group, with no matrix at all, picked the losing project 44.7 percent of the time.
The bread nobody talks about
Imagine you run a regional bakery. Its fast-growing gluten-free line gets the attention at every meeting. The wholesale bread you sell to cafés holds a small share of a flat market. On the matrix, it’s a Dog.
Two requests reach you in the same week. The gluten-free team wants a marketing push. The wholesale manager wants a new oven that would cut energy and labor costs on every loaf. You have money for one.
Now set the matrix aside for an hour and lay out the cash each option is expected to bring back, year by year, against what it costs. If the oven pays back in three years and the campaign’s numbers keep shrinking after the first, the box the wholesale line sits in doesn’t change the arithmetic.
You may still back gluten-free, for reasons the numbers don’t hold. Then at least the reason has a name, and it isn’t “Dog.”
Defending the Star
Armstrong and Brodie surveyed MBA students too. Of 175 who answered, two-thirds believed the matrix would produce better decisions under certain circumstances. They couldn’t describe the circumstances, and most said their evidence was what the textbook or instructor had stated.
In the experiment’s debriefing sessions, the administrators reported, people defended their decisions vigorously.