Monthly Archives: January 2013

Analytically Competitive Companies – The Key Missing Link

It is all about “What is your strategic measurement?”

What is not measured can not be managed and what is not understood as strategic measurement is the path to the death of an organization.

“search dominance” is Google’s strategic measurement, not how many androids were sold, nor the stock price of the company.  So how does that fit in the ‘eWallent’?  It depends on how it is implemented.

“Functional and beautiful consumer computing platform” is Apple’s strategic measurement, not really making maps or replacing Youtube application.  It was a double whammy when they failed in maps, because market understands that when Apple is doing something that does not fall under their strategic measurement and in addition to that they fail miserably in the quality.

Interestingly strategic measurement is also very tightly connected to the vision and mission of the company.  Boy!, how many companies thought that exceptional customer service or ‘bill of rights’ as the strategic measurement to follow and hang those CEO signed boards around the organization!

Davenport and Harris, in their rock star analysis of how to build a analytically competitive companies in “Competing On Analytics – The New Science of Winning (2007)”, provide a very detailed structure of five stages of analytical companies identify the stages as

– Analytically impaired (Stage1)
– Localized Analytics (Stage2)
– Analytical aspirations (Stage3)
– Analytical companies (Stage4)
– Analytical competitors (Stage5)

The key set of attributes of these companies are included as a tabulation in Table 6.2 in the book, where the attributes that differentiate these stages for the level of understanding and application are listed as

– Analytical objective
– Analytical process
– skills
– Human sponsorship
– Culture
– Technology

I want to add to this list the importance, understanding, and application of “The Strategic Measurement”.

The case study I will bring to bear these points is the Moneyball.

The key strategic measurement that convinced the GM of Oakland A’s is that OBP matters not Slugging.  See the detailed discussion on these concepts in http://predictive-models.blogspot.com/2012/07/analytical-success-path-is-moneyball.html

I see understanding strategic measurement and relentlessly pursuing it requires a strong commitment and it can not happen unless you are very sure of it supported by years of thought leader’s intuition, business acumen, and subject matter dynamics all supporting the relationship between strategic measurement and operational success from different angles, including key observational model in support of likelihood of success.

The challenge is ‘do you know your strategic measurement’?

So how is it related to the five stages of analytical competitiveness?

From Data Monster & Insight Monster

A Simple Approach For an Optimal Design For a Practical Problem

Let us say we have a design to test in a digital world, optimizing what matters in our web -design and at the same time test various other possible consumer states such as

4 Types Offers

4 Display variation of the link, among

5 Segments, and with

4 price level, resulting in a total of 320 cells.  

However, in practical life, we will always test the main effects which is in this case testing 17 possible main effects, and we have to test the two way interactions and we do not want more than two-way interactions in our analysis.  If you end up doing all-way interactions, with the intention of estimating all possible class estimates – that is we do not use a common mean for the left out classes – then we will end up using

There are many ways to approach this practical problem.  While the well studied path is the classic design of experiments looking for latin square, greeco-latin square, super latin square, fractional factorial analysis and so on.

The  purpose here is to use the regression approach to solve our problem.