Light entertainment: volume visuals
From text documents to our eyes

Reading behind the chart

I could have filed this one under Light Entertainment but it's too good a chart to lay to waste:

(The chart is from Internet Retailer.)

Let's focus on the (mis)match between the question being addressed and the data collected to address it. The intention of the analyst is fully divulged in the title of the accompanying article: "Don't sell Twitter short: those 140-character messages reach an affluent and engaged audience". The chart supports this claim by showing that Twitter users disproportionately represent the types of consumers that marketers most covet, i.e. those with advanced education, and those earning higher incomes.


Up top, the concept "user" is very pliable. Is it someone who has a Twitter feed or is it someone who reads Twitter feeds or is it someone who subscribes to Twitter feeds? Is it someone who is registered or everyone who visits? Is it someone who has visited the site in the last x months? or posted a tweet in the last x months? If a writer, does it include someone with no subscribers? no page views? What about people who simultaneously publishes multiple feeds (like John Cook who writes one of my recommended blogs)?

Now, we don't expect the analyst to describe fully how a "user" is defined but while interpreting this chart, we should ask appropriate questions of the data.

Next, the analyst establishes a reference level called "general population". Is this the right metric? This depends on what the chart is used for. If you are choosing between spending money on Twitter, and say spending the money on national TV advertising, then perhaps this comparison is valid. If you are selecting between Twitter and say Google, then absolutely not. For most readers, I think a more relevant point of comparison would be the general Web user, rather than the general population. This is an important distinction because the general Web user also earns higher incomes and has higher educational attainment than the general population, thus the current set of data exaggerates the "value" of Twitter exposure.

Finally, if you are a marketer looking to spend with Twitter, you are also worried about "reach". Say, 50% of website ABC's users are rich people compared to 10% of your reference population. That sounds like an amazing opportunity. Well, only if website ABC has sufficient number of users! If ABC is a niche website serving only 1% of your reference population, then despite the benefit of targeting, its scale is too small for your need.

Don't take a chart on its surface. Read behind the chart!




Feed You can follow this conversation by subscribing to the comment feed for this post.

Mark Thomas

Not to mention that the circles suffer from the classic mistake of using diameter rather than area, further exaggerating the results.

The comments to this entry are closed.