Category — Statistics
What happens when good data is used the wrong way? The data gets all the blame, as seen in the latest round of finger pointing about ComScore and their reporting of Google click growth.
In February, when ComScore, a market research firm that tracks online marketing and advertising, published findings that found growth in Google’s paid click traffic was down significantly versus the prior year, Google’s stock fell into a steady decline, retreating almost 45% from its 52 week high.
However, when Google announced recently that its 1st quarter profits were up 31% versus the prior year, Google’s stock shot up, while ComScore’s own stock was pummeled. Google’s CEO Eric Schmidt even made a veiled reference to “third party” data missing the boat when it came to predicting Google’s performance.
The problem was that what ComScore was measuring (Google’s paid click growth in the US) wasn’t matching what Google was reporting (global and third party paid click growth). According to Andrew Lipsman at Comscore:
“The main difference between the paid clicks trends reported by Google and comScore can be traced to the fact that the comScore paid click data cited in financial analysts’ reports (and subsequently reported by the media) are U.S. data only.”
Additionally, when you do an apple to apples comparison, the trends between ComScore and what Google reports itself are directionally the same:
Not surprisingly, the nuance of research data can generally be lost on the investment community.
However, just because data is used in a wrong way, doesn’t make the data wrong itself.
April 28, 2008 No Comments
- Germans tend to comment more on social media than participants in other EU countries, but not half as much as those in urban China.
- People in the UK are not as big on tagging or using RSS feeds as others, but they lead Europe in social network participation.
These were just some of the factoids I uncovered while having some fun over the weekend playing around with this social media segmentation tool from Forrester Research (I’ll be the first one to admit that only someone who works in market research would call playing with social media segmentation tools a form of fun).
The tool looks at broad groups of internet users and classifies them into segments based upon their level of social media participation. This chart shows the different groups and their definitions (click for a larger image):
My only quibble with the segmentation is that I believe there is a lot more granularity out there for Creators (e.g., I’d imagine there could be some significant differences between those who strictly blog versus heavy YouTube or Flickr uploaders).
But the segmentation does do a good job of grouping the population as a whole on something more meaningful than demographics in order to make better sense of them. Maybe, however, there is a micro-segmentation of Creators lurking somewhere under the hood of Forrester’s research.
Here are some other facts that Charlene Li mentions on the Groundswell blog that you can find if you twiddle around with this tool:
- Although social media participation significantly increases the lower on the age range you go, even among the 55+ group in the US, you’ll find that 33% of them are connecting with social applications in some way.
- 41% of Koreans are Joiners — members of social networks — more than anywhere else in the world.
- In Urban China, a full 36% are Creators, which means that this very significant percentage of the population is creating blogs, maintaining content, or uploading videos or music.
I’d like to play around more with this dataset (hint to Forrester!) and maybe later I’ll have some more insight nuggets to share.
March 31, 2008 No Comments
This post by Seth Godin on using survey results in marketing got me thinking about one of the most memorable survey results in market research history, the one he references from the old Trident gum commercials:
“Four out of five Dentists surveyed recommend sugarless gum for their patients who chew gum”.
While it sounds impressive (which is why everyone still remembers it), the sharp eye will see the potential push behind the fact. The key, as with most published statistics, is in the qualifiers, this time in the sample base and around the word “recommend”.
On the sample base, it’s a potential signal that it may not represent all dentists, just the ones who were “surveyed”. If you have a large and truly representative sample, why would you be using that qualifier?
The second is the qualifier for the word “recommend”. The dentists surveyed aren’t recommending that you chew Trident, just that if you do chew gum, it should be sugarless. No one asked whether you should be chewing gum in the first place or if any other brand of sugarless would suffice. Like percentages in my previous post, there can always be fun with numbers.
November 2, 2007 2 Comments