Big Data

Model your analytics approach on a statistical normal distribution

John Weathington points out interesting correlations between normal distributions in statistics and informal norms as they're distributed throughout analytic organizations.

Image: Picasa

If your corporate strategy embraces big data and analytics as a way to build analytic capability within your organization, there are interesting ways you can use a statistical normal distribution in modeling your approach to analytic excellence.

Normal distribution's distinctive characteristics

Perhaps the most obvious feature of a normal distribution is its distinctive characteristics. There are statistical tests for normality (Pearson's Chi-Squared, Anderson-Darling, etc.); however, you can usually tell if a distribution is normal just by graphing it.

Bell shape

Normal distributions are symmetrical and carry a distinctive bell shape; in fact, when people refer to the bell-shaped curve, many of them don't know they're referring to a normal distribution.

Similarly, organizations with a high degree of analytic capability have a distinctive appearance. When I go into an organization to initially assess analytic competence, one of the first things I do is walk the halls — I observe what people are doing and what they're saying. I can very quickly get a sense for what I'm dealing with, because analytic capability is very observable. People talk about confidence intervals, variation, and statistical significance. Charts and graphs line the halls, and huge monitors display critical scorecard data in real-time.

Clean, simple, and perfectly aligned

The shape of a normal distribution is not only distinctive, it's also clean, simple, and perfectly aligned. (To contrast, a chi-squared distribution looks lopsided.) In addition to being symmetrical, everything about a normal distribution's central tendency is consistent. In statistical terms, its mean, median, and mode are all the same. It's not important to understand the exact differences between these three measurements — what's important is that they all measure a particular quality of any distribution (central tendency), and in a normal distribution there's no difference between the three.

A truly analytic organization has the same consistency in culture. Many organizations behave differently from their stated beliefs. That's why it's so important for me (as a management consultant) to observe what's happening within the organization and compare that to what top leadership says about the organization (e.g., in an annual report). When there's a disconnect, it's important to run a cultural intervention so I can unearth assumptions that are driving behavior that's inconsistent with what the leaders in the organization want. When an analytic organization is properly installed, all measures of cultural orientation are aligned.

A prerequisite for success

Normal distributions are a necessary requirement for other types of analyses. This is an area that often trips up novice statisticians. For instance, a control chart is a popular way to measure the quality of a process; however, some of the analyses in a classic control chart assume that the underlying distribution of the process data is normal. Applying standard control chart techniques to a non-normal distribution will trigger false alarms.

Similarly, it's difficult to control an organization by its scorecard if its culture doesn't appreciate the value of information and analytics. I've often seen leaders jump enthusiastically into building a corporate scorecard with the expectation that the rest of the organization will see the light once the data becomes visible. Disappointment sets in when nobody really cares about the visible disconnect between target and actual. An analytic culture is a prerequisite for strategic success, not a destination.


Normal distributions in statistics give leaders a good metaphor for analytic competence. Analytic organizations have the very recognizable qualities of being fully aligned between stated beliefs and observed behavior, leaving no area for underlying assumptions that could compromise organizational objectives.

Also, if your plan is to leverage big data analytics for strategic success, you must make sure you have an analytic culture in place to support it. Take some time today to assess the mindset and behaviors that are currently driving your company — it may be time to distribute a new norm.

About John Weathington

John Weathington is President and CEO of Excellent Management Systems, Inc., a management consultancy that helps executives turn chaotic information into profitable wisdom.

Editor's Picks