I recently ran a call tracking campaign for a well known national brand. In this campaign, users see an ad on a mobile site or app. Clicking on the ad prompts a call to the advertiser's call center. A call resulting in sale is counted as a conversion.
This particular campaign ran in 42 cities in south US. Users in each city saw the same ad, on the same devices and on the same set of sites and apps. After running for two weeks the results were intriguing. Conversions varied widely from city to city. People in Orlando and Florida were 10 times more likely to respond to the ad and call the call center than people in Nashville and Indianapolis.
In the heat map below each city targeted is represented by a circle. A large circle means high conversion rate.
It sure looks different but could it be just a coincidence?
To find out I ran a Chi-Square for statistical significance. It refutes the hypothesis that this result set comes from a uniform distribution at confidence level of 0.9995. For those who skipped statistics in college, it can't be a coincidence.
If you can't see this map properly,and you are using Chrome, try opening the page in Firefox. Or you can view it here.
This particular campaign ran in 42 cities in south US. Users in each city saw the same ad, on the same devices and on the same set of sites and apps. After running for two weeks the results were intriguing. Conversions varied widely from city to city. People in Orlando and Florida were 10 times more likely to respond to the ad and call the call center than people in Nashville and Indianapolis.
In the heat map below each city targeted is represented by a circle. A large circle means high conversion rate.
It sure looks different but could it be just a coincidence?
To find out I ran a Chi-Square for statistical significance. It refutes the hypothesis that this result set comes from a uniform distribution at confidence level of 0.9995. For those who skipped statistics in college, it can't be a coincidence.
If you can't see this map properly,and you are using Chrome, try opening the page in Firefox. Or you can view it here.
This comment has been removed by a blog administrator.
ReplyDelete