Although not formally released by Harvard Business Press for another week (Amazon.com indicates February 8, 2010 availability), as expected the success of "Competing on Analytics: The New Science of Winning" is generating interest in the "Analytics at Work: Smarter Decisions, Better Results" follow-up effort by Thomas H. Davenport, Jeanne G. Harris, and Robert Morison. The first several reader reviews have been submitted (including my own review), and I had an exchange with one of the first reviewers, Mark P. McDonald, in response to one of his comments, reiterated here:
Mark P. McDonald:
While the book discusses analytics at all levels, it tends to concentrate analytics activities into a specific group of subject matter experts. While I agree that analytics requires specific skills, setting these `quants' up as a special group may limit the spread of analytics across the enterprise. This is a minor point that does not reduce the value of the book.
Erik Gfesser:
Regarding your comment: "While the book discusses analytics at all levels, it tends to concentrate analytics activities into a specific group of subject matter experts. While I agree that analytics requires specific skills, setting these `quants' up as a special group may limit the spread of analytics across the enterprise. This is a minor point that does not reduce the value of the book." Based on my reading of this book, I need to disagree with what you are saying here. The authors repeatedly state that the spread of analytics in any given firm depends in part on the incorporation of analytical thinking across the enterprise, rather than limiting analytics to the quants. In addition, the authors point out that the abilities of quants need to be at a level far above what can be found in simple readings of a hypothetical "Analytics for Dummies" genre of texts - the knowledge required to do the heavy lifting to which the authors refer is not trivial due to the need to have a deep understanding of both statistics and the associated industries in which quants work. The philosophy that the authors expound on this subject is much akin to the enterprise adoption of Six Sigma and other process improvement or quality initiatives in the sense that limiting the knowledge of the benefits that such initiatives provides and the necessary steps to achieve associated goals to just Black Belts would jeopardize progress.
Mark P. McDonald:
Erik you are right and I mentioned this issue as one of the few about only weakness I could find. So agree with the comment and share your enthusiasm for the book and its ideas. But one point, you liken this to Six Sigma and I can see the value in the analogy. However one of the things that I have always admired about six sigma is its deployment model -- its ability to create green belts, black belts and master black belts out of every day people in the company. That deployment model should exist as well with analytics, particularly since SPC is a big part of the Six Sigma tools.
I will reiterate my recommendation for this book.
Erik Gfesser:
Interesting points. While statistical process control (SPC) might very well fall within the realm of analytics, depending on definition, Six Sigma is very much concerned with quality. The definition of "quality" can also vary depending on the project - e.g. the term might mean that the cycle time for a process needs to stay below a set limit to provide a predetermined level of cost savings - but Six Sigma is really all about reducing defects to provide measurable business value. The tools to achieve such quality goals are numerous, and many require in-depth statistical knowledge to use even if software packages might be available to aid the Six Sigma practitioner during project execution. However, the goals of the analytics in the discussion here are varied - i.e. even though Six Sigma is far from being a trivial process improvement methodology, the scope of "analytics" is much broader. The authors indicate that their definition of analytics is "the use of analysis, data, and systematic reasoning to make decisions", and further their explanation as follows: "What kind of analysis? What kind of data? What kind of reasoning? There are no hard-and-fast answers; we contend that almost any analytical process can be good if provided in a serious, systematic fashion". The analyst equivalents to the various Six Sigma Belts and Champions might vary well be what the authors refer to as "amateurs", "semi-professionals", "professionals", and "champions", but the universe of analytics is simply not as narrow as Six Sigma, and can therefore be harder to reproduce in the fashion you mention.
