Power to the People: Analytics for the Masses

By Neil Raden

In an article in the January/February 2006 issue of the Harvard Business Review titled “Competing on Analytics,” Tom Davenport makes a case for the use of advanced analytics as a competitive differentiator. I should add that the funding for his research came from SAS and Intel, but there is a disclaimer that “this research was carried out independently.” For those of us involved in data warehousing and BI and even, like myself, an old number cruncher from way back, this might sound very encouraging, but there are some alarming conclusions in this paper.

Davenport, who is quite prolific on the topics of business process reengineering, knowledge management, and, in a recent issue of Harvard Business Review, the idea of business process commoditization, is really out of his depth on this topic. His central premise is that companies are starting to employ analytics (which he defines as statistical and predictive analytics) as the primary elements of competition, replacing intuition and gut feel. To support this claim, perennial poster-companies like Wal-Mart and Harrah’s are showcased. These organizations may have a long history of using predictive analysis to streamline and enhance their businesses, but these examples fail to convince that there is any great movement in this direction.

In an exchange with Davenport, my argument was met only with, “Where are your numbers?” Twenty-five years of experience is, in my opinion, more weighty than a survey. Nevertheless, I did compile some numbers, about Davenport’s group. As is turns out, the group has been underperforming the stock market:

 

 

 

 

 

 

 

There are two schools of thought when it comes to the value of BI in general. One is that it is best used by “quantitative” types and other analytical business people, who can spot trends and analyze patterns to assist in the big decisions and set and direct strategy. The other position is that BI is at its best when helping a broad range of people and processes at an operational level, marginally improving performance, repeatedly and often. The former is the commonly held view of management consultants and, previously, BI practitioners a decade ago. The latter position gained currency in the last few years and is now widely seen as borne out in practice. Using BI to form a new strategy for a global financial services firm makes for good marketing collateral, but when it comes to ROI, lots of small improvements are the way to go.

Davenport is firmly in the first camp. In fact, his argument is based on three principles:

Analytics requires Ph.D.s. The scheme he recommends is a centralized team of quantitative experts who support this function for the entire enterprise.

Success depends on commitment from senior executives. The implication is that unless the senior staff is behind the idea of advanced analytics replacing intuition, gut feel, common sense, and industry experience, it doesn’t have a chance.

Centralized control of data and expertise. One section reads, “The difficulty is primarily in ensuring data quality, integrating and reconciling it across different systems, and deciding what subsets of data to make easily available in data warehouses (emphasis mine).”

Taking these one at a time, there is actually a mention in the paper of one company that recruits “Ph.D.s with a personality.” This kind of condescending feeling is pretty common. Years ago, when companies kept a group of statisticians, mathematicians, and operations research people in a unit to work on the hard problems, this attitude was pervasive—that people who could do this work were not really cut from the same cloth as, say, the VP of marketing or the controller. I can see no reason why this would be any different now, especially with comments like the one above, so it is hard to fathom how organizations would suddenly turn their decision making over to experts. In fact, when it comes to quantitative modeling in business, there is a recurrent paradox—the more complex the model, the less faith people put in it. People take advice from people like themselves; that’s why there are so many lawyers in government and so many finance people in the chief executive’s office.

This old theme of the operations research department that didn’t work throughout the 1970s, 1980s, and 1990s is just not tenable. Besides, analytics today can be dialed up or down depending on the audience. It can be made understandable to domain experts. Analytics doesn’t have to be difficult, and there are many cases, some with companies mentioned in Davenport’s paper, where very sophisticated analytics are being used by marketing, sales, purchasing, and finance departments, because the tools have been made more accessible (and, of course, the data is being provisioned, eliminating the need to find and clean it). The best analytical tools now incorporate real-time interactive visualization, a topic Davenport missed completely. I’m going to send him a copy of Spotfire.

There is one area where Davenport’s vision tracks well with current practice. Centralized data mining/predictive modeling groups are capable of discovering valuable insights that can then be encapsulated into reusable algorithms, scores, or rules. Everyone is familiar with the Fair Isaac credit worthiness score. It isn’t necessary to understand the model behind the score to be able to apply it. In the same way, specialized data modeling groups can produce useful tools for others with a wide range of skill. However, this by no means implies that the most valuable or useful analysis should occur solely in this group.

Senior executive support is always a good thing to have, but I can find no support for the idea that it is crucial for the adoption of analytics in a company. Of course Jeff Bezos supports it—Amazon is a 100 percent Internet business and was formed in a time when using data to run your business was accepted. But a 100-year-old candy company, whose business is pretty much the same every year, is going to spend much more time looking at supply chain and fulfillment reports than using Markov chains to decide which mix of flavors works best in the Christmas promotion. Clearly, what Davenport is getting at is the conversion of organizations from current decision-making processes to using predictive modeling up and down the corporate hierarchy—but that is neither likely nor necessary in most organizations.

Centralized control of data and analytical expertise may not seem very controversial, but what Davenport is implying is not only centralized control, but also centralized design. This is another naïve assumption, because many organizations are not only decentralized—they’re dysfunctional. Separate units within organizations often need autonomy because they are just so different from the rest of the organization. In addition, as an organization becomes more “agile,” which is a definite trend, decision-making, even for the big decisions, will become more decentralized. Imagine how difficult it will be to buy or sell pieces of a company if the “brain,” the centralized analytical capability, stays with the parent and there is no local expertise?

It may not be glamorous and it may not reek of the “next big thing,” but little bits of BI attached to the smallest processes and process steps seem to have enormous impact and potential for continuous improvement. Thought leader Peter Drucker has said that it was important to find a way to run organizations with the honest contribution of ordinary people, not the efforts “of a few supermen.” Davenport’s vision of a centralized cadre of mathematical geniuses making all of the important decisions in a centralized organization is neither feasible nor desirable.

Neil Raden is the founder and president of Hired Brains (www.hiredbrains.com). He is an industry analyst and working BI practitioner. He can be reached at nraden@hiredbrains.com, and welcomes your comments.