cover story

Data: The CEO's Drug of Choice

By Kelly Shermach

In the early 1990s, PepsiCo developed a clear cola. Focus groups in the United States liked Crystal Pepsi, and its launch in Europe and Australia was successful with consumers repurchasing the product and making it a part of their thirst-quenching pantry.

Pepsi moved the product to U.S. test markets. There, however, it did not receive positive results. It did not hit rollout requirements nor achieve consumer repurchase, the telltale sign of a consumer packaged goods winner.

Still, Pepsi introduced the product across the nation. “There is so much data in consumer packaged goods because the data is collected by infinite categories—geography, size, color, flavor, scent, etc. If you go through enough data variations, you can predict almost anything,” says Rip Gerber, chief marketing officer at Intellisync Corp., a software developer for personal digital assistant products (PDAs). But for self-preservation, executives must look at the forest, not just the trees.

U.S. marketers substituted the poor results from American market tests with the data proving successful in Europe and Australia. For good measure, the marketers revamped their marketing plans, replacing the tune focus groups had heard with Van Halen’s “Right Now.” There was market data that testified to the products’ profitable future, and it was all the team and Pepsi’s CEO needed to ramp up production and media buys (including SuperBowl spots, a promotional CD anthology of popular music and some of the first-ever bus wraps).

The data that sold Pepsi executives on the U.S. product launch also sold them out. Like white powder on the upper lip of a cocaine addict, the data from overseas markets gave away PepsiCo’s addiction to data. A lightly carbonated beverage that tasted like cola but was as clear as water did not entice repeat purchases and resulted in a very public firing of a Pepsi executive.

“One way to fail is to always answer the CEO’s question asked rather than say, ‘That’s the wrong question,’ or ‘Ask it this way,’” says Gerber. “Sometimes you have to push back.”

Of Ignorance and Hubris

Companies are overflowing with data —so much of it that they sometimes rather clumsily harvest it to conceive of huge changes or justify ill-conceived decisions.

In some instances, the haphazard use of data is an issue of ignorance. Business majors do not receive much education on, nor have corporations placed a priority on data analysis. Rather than submitting data to a business intelligence tool, executives look to their subordinates to give the data warehouses a good glance. At times, the CEO seeks data that supports a pet project he would like implemented or something that reinforces a gut feeling about market desires.

Both of these approaches more often than not yield products or business changes that do not work. Data becomes not the enabler of success but the rope with which executives hang themselves.

“The more data you give an executive, the better they feel about their decision,” says David Silverstein, president and CEO of Breakthrough Management Group, a Six Sigma-focused consultancy in Longmont, Colo. Silverstein suggests HP’s acquisition of Compaq Computer as an example. When HP announced that it was acquiring Compaq in 2001, both companies’ stocks took a hit. The merger offered no obvious benefit to the companies, their customers or the information technology industry.

“None of the reasons HP’s acquisition of Compaq is turning out to be successful were in the original objectives,” he says. Both companies were suffering from a protracted weakness in the hardware market and sought to cut costs. They also wanted to strengthen the ability to compete with IBM and depended on merged marketshare to make them formidable in the face of Big Blue.

Yet there was so much overlap between HP and Compaq product lines, the combined pair could not dwarf IBM in market share measures. It took years for HP and Compaq to begin to make something valuable together. In the meantime, revenues and marketshare dropped more than