Beware the Hidden Risks in Business Analytics
There was no way that competitors were a serious threat to Nokia at the turn of the 21st century. Its revenues and profits exceeded $20 billion and $2.5 billion respectively and 55,000 employees were making and selling mobile phones that dominated the markets of Europe, Africa and Asia. The once aggressive Motorola was fading. No competition came from Google as it concentrated on search engines. Samsung was a dull Korean industrial complex.
Thus, we have one of the earliest examples of the misunderstanding of business analytic risk. Apple’s quirky Steve Jobs paid little attention to flawed data that forecast a limited market for smartphones. His success is an early warning sign of opportunities and risks arising from big data.
By way of background, business analytics can be traced back to 1908 and Henry Ford’s Model T automobile. Data was analyzed to help Ford build low-cost cars offering durability, versatility and ease of maintenance. Today, companies collect and analyze massive databases to obtain 21st century competitive advantages. Everybody’s in the game.
A recent count shows 137 U.S. colleges and universities offer degree programs in business intelligence, data science or business analytics. In all the programs, high-level programming languages, regression analysis and other powerful tools mathematically and visually describe relationships among independent and dependent variables. If we know causation or correlation, we gain a better understanding of how to take advantage of current trends and changing markets.
To win the race, we may need to pay more attention to the driver.
All of this is good until we introduce risk management. Many schools are taking a narrow mathematical approach to big data. They may not be preparing students to interpret data in the context of business processes and market behavior. Their students may not be encouraged to recognize that analytics produce more accurate results if they are developed with a curiosity about changing non-quantitative trends and behaviors.
Evidence to support this contention comes from factors including:
- Massive Systems. We are linking data from financial, health care, credit cards and personal behaviors into a single database. The potential is enormous for complexity to destroy data validity and reliability.
- False Accuracy. Quantitative tools often overpower the data. If we multiply 4 times 76.34715, we cannot claim accuracy to the level of five decimal points.
- Limited Perspective. We do no one any favors teaching business analytics without a framework of business, psychology and critical thinking.
Identifying strategies based on prior statistical relationships is only valid when the analyst considers how current events might be changing the data. Absent this recognition, business analytics is making predictions about the past, not the future.
Should we be worried when business analytics is analogous to a race track where owners are going ’round and ’round while constantly seeking new ways to outwit competitors? In this context, regression analysis is a Ferrari. The problem is not the vehicle. To win the race, we may need to pay more attention to the driver.
From a risk management perspective, schools should ensure programs in business intelligence and data science contain content beyond high-level programming languages and complex quantitative techniques. All work and no play makes Jack a dull boy. All regression and little understanding of the context of business and changing markets produces results that can be horribly misleading.
Just ask the energetic and talented managers who were forced into early retirement at Nokia.