What are the data sources?¶
The vast majority of the data are geotagged twitter and instagram posts.
What is the time window?¶
The data is based on a running 24 month time window. This timeframe accounts for seasonality and tends to be the most predictive of retail sales.
How is this different from MOSAIC, PRIZM, etc.?¶
Geodemographics such as MOSAIC and PRIZM are a classification of areas based on demographics of the people who live in that area. By applying survey data such as GFK MRI analysts get a sense of what behaviors to expect from an area based on the area’s demographics.
Our Proximity dataset scores areas for various attributes based on what people are doing and saying on social media in that area.
In this case two areas can be similar demographically yet have different social behaviors that can impact sales and as a result lift a predictive model.
Is social media mainly used by youth?¶
15% of social media users are over the age of 55*. There are specific segments that correlation with age: such as “student life” where the age is 23% below that national mean, or “farm culture” where the age is 9.79% older than the national mean.
How is volume accounted for?¶
We provide a volume index that is directly proportional to the amount of posts in an area. This can be used in a weighted average for a trade area.
Do the posts come from where the behavior was tagged or the home block group of the user?¶
Posts come from where the behavior was geotagged. For this reason a tighter trade area is recommended to reveal what is happening near a store or in a area.