Marketing Technology for Growth

The Whimsical Nature of Data Analytics

the whimsical nature of data analytics

The other day I came across two articles that, when taken together, aptly underscore the fickle and sometimes outright contradictory nature of marketing data: In-Store Gets the Cold Shoulder, as More Women Favor Web Buying, and We Love Amazon, But Going to the Store Matters Most. After reading both headlines, I was left asking one simple but profound question: “well, which one is it?” In fact, the contrast between the two only got starker as I read on…


The first piece, written by reputable marketing-tech analytics firm eMarketer, based its analysis on the findings of a survey of over 2,000 US female internet users in March 2013 conducted by Women's network SheSpeaks and female-oriented marketing firm Lippe Taylor; in the survey, only 6% of respondents reported still researching products primarily in-store; the majority of the rest (89%) did their browsing mostly on the web, either via desktop (71%) or on mobile devices (18%).

From these numbers, the eMarketer article drew the following conclusion: “there is no question that for retailers trying to reach women early in the purchase path, online is the place to be.” Seems sensible enough; OK, fine, got it.


However, then I read the second article, this one from equally reputable The Wired post based its analysis on a recent survey conducted by Forrester Research of 4,500 U.S. adults online, which found that in every major consumer category other than travel, shoppers said visiting a store served as the most important source of research before buying.

Now wait just a minute here…


Reading these two articles got me thinking, how does one go about reconciling seemingly discrepant data, anyway? Remembering that the devil is often in the details, I assumed clarification could be found by analyzing the parameters of the surveys in greater depth. I started with the Forrester one. According to the article, it involved “4,500 U.S. adults online.” Hmmm, does this mean they surveyed 4,500 adult online users or 4,500 adults though online means. I wasn’t sure. The one thing I did know was that I was too cheap to pony up the $499 Forrester was asking to access the full study and clarify this question. Sorry, gang – a blogger’s salary ain’t what it used to be.

Absent a detailed understanding of survey methodology, I thought I could turn to deductive reasoning. Because the SheSpeaks study indicated such an overwhelming connection between women and online research (89%), while the Forrester study suggested a strong correlation between US adults and in-store research, one could deduce that somehow Forrester forgot to include women in its sampling. However, with roughly half of the US population being female, and a sampling size of 4,500 respondents, this seemed a fairly unlikely oversight.

Further commentary from the post hinted at a darker possibility:

“In all, Forrester’s findings suggest no one in the selling business can afford to ignore the primal satisfaction of touching holding something in your hand before you buy. Human toolmaking and trade both started as hands-on endeavors…”

Given this excerpt, I was tempted to conclude that Forrester does not regard female online users as human. This one seemed a bit of a head scratcher, though. After all, I presume there are plenty of women employed within the Forrester ranks who might raise an eyebrow at such exclusionary methodology; I know many women, my wife for one, who would surely take exception.


All silliness aside, this tale of two disparate data sets reveals perhaps the biggest challenge big data poses for businesses and marketers – how to glean accurate, relevant and actionable insight from the numbers.

A really interesting article on the science of data recently published in FastCoLabs touches on this very issue. In the post, author Ciara Byrne quotes data scientist Jake Kyamka as saying, “Data science is not just about number-crunching… It's all about people. The data comes from what people are doing, great data scientists have an ability to understand people and the ideal result is something which is going to help people.”

This seems sensible enough. In the examples cited above, we have two discrete data sets from which two radically different conclusions have been drawn, each meant to help a certain subset of people glean meaning from data, usually to bolster their pre-conceived notion, argument, or perspective.

As such, the devil is not in the data, but rather in how it is sliced and diced, how it is interpreted and relayed.

In an era rife with data, the greatest challenge for businesses and marketers lies not in the data itself, but in its meaningful interpretation.




Topics: Data Analytics