November 5, 2012
With one day to go until the election, all of the headlines about red states, blue states, and swing states might have you feeling like you’re in the middle of a spin art painting. But when it comes to ecommerce, there’s good reason to keep examining state-by-state trends in politics and shopping habits.
Case in point: Monetate engineer and data guru Jeff Patti combined state-by-state election forecasts from the New York Times’ FiveThirtyEight blog with Monetate’s own data from more than 300 million ecommerce shopping sessions to uncover some fascinating trends.
“I took the data from the projected election percentages, specifically which percentage of a state is likely to go Obama, and correlated it with the share of traffic to leading ecommerce websites from people based on the platform they were using, their referral URL, and their browser,” Patti explains.
A correlation is the relationship between two variables in a population. The question: As one variable increases, is the other variable also reliably altered? A trivial example would be height and pant inseam size. A taller person is going to consistently need longer pants, which would make this relationship a strong correlation. Also, this example would be positively correlated, since an increase in height would generally yield a corresponding increase in pant length.
In his correlations, Patti was able to research whether iPhone users were likely to vote Republican or Democrat. It turns out there is no correlation between the device and political party: Both red states and blue states are just as likely to suffer from iPhone addiction.
But the most striking finding in Patti’s data, particularly for marketers, involved picture-happy behemoth Pinterest.
“The Pinterest correlation is certainly the most impressive and surprising,” Patti says. “The more people that come to our websites from Pinterest in a state, the more likely it is to be a red state. Pinterest is negatively correlated with being democratic as a state.”
That’s right: States that have more people following links from Pinterest to online retailers are more likely to vote for Mitt Romney. In Patti’s research, this was the strongest correlation.
It’s perhaps an unexpected finding, given that most people would assume Pinterest attracts more artsy, liberal users, or that there would be no political correlation at all since it’s such a popular social media network.
Patti’s research is intriguing for a few reasons. First, as we’ve already covered in our election day infographic, ecommerce spending trends by political party can reveal new segmenting opportunities. Second, Patti points out that this data proves again just how important it is to examine timely trends in website traffic.
“It just goes to show you that targeting on a state-by-state basis is important because there are very surprising correlations sometimes,” Patti says. “Using a platform that allows you to change your messaging on a state-by-state basis might help promote your brand or improve sales because there are major differences between states.”
Bottom line: It’s always a good idea to test assumptions about what website visitors want. But it’s also a good idea to test assumptions about traffic based on location and political affiliation in order to target effectively during an election season.