Skip to content

Dataset Structure

The FollowGraph dataset provides quantitative measures of likelihood to follow influencers, brands, celebrities, media outlets, interests, and more via social media data.

Index Scores

FollowGraph expresses likelihood to follow as a 100-based index relative to the national average for any variable.

\[ Index \, Score = \frac{Predicted \, Percentage}{National \, Percentage} * 100 \]

Let's say that, on average, 2% of Americans follow the NFL on social media. However, we have predicted that the likelihood of given person (or people in a given area) to follow the NFL is 4%. This person (or area) would receive an index score of 200 to indicate that they are twice as likely as the average person to follow the NFL.

\[ \frac{4.0 \%}{2.0\%} * 100 = 200 \]

Individual vs. Geographic Level Data

As has been previously noted, FollowGraph can quantitatively score follow likelihood for individual people as well as aggregate follow likelihood of the people who live in a given area. Individual level data can be more precise and targeted for marketing and other use cases. On the other hand, area data is a useful tool for making location intelligent, data-driven location decisions.

Here's an example:

For the "Los Angeles Lakers" variable, a neighborhood in LA scores 800. This means that on average, the people in this neighborhood are 8x more likely than the average American to follow the Los Angeles Lakers on social. However, in addition to where they live, variables like an individual's age, income, gender, ethnicity, and more are important factors in their likelihood to follow the LA Lakers. The individual people in the neighborhood range from 4x to 14x as likely.

Variables and Categorization

FollowGraph variables are organized using a two level categorization. In the first level, all variables are categorized as one of the following:

  • Brands - Businesses and brands with a social media presence. Typically, consumers can explicitly purchase goods or services from these companies (more info below).
  • Channels - You can think of these like marketing channels. What are the podcasts, shows, TV channels, online magazines, etc. that consumers follow?
  • Interests - Interests based on likelihood to follow a combination of accounts. You can learn more in the methodology section.
  • Celebrities and Influencers - Celebrities, influencers, politicians, athletes, and more.
  • Organizations - Organizations like sports teams, government organizations, and more don't always fit neatly in our brand definition (see below).
  • Entertainment - Entertainment related accounts that fit are more aptly categorized here
  • Miscellaneous - There's a lot of variety in social media. Many popular accounts and groups of accounts don't fit the other categories.

Each of these categories is further broken down into subcategories.

  • Brands - Auto, Beauty, Entertainment, Fashion, Finance, Fitness, etc...
  • Channels - Magazines (online typically), News Publishers, Podcasts, Radio, TV Channels, TV Shows, etc...
  • Interests - Apps & Tech, Automotive, Business and Tech, Fashion, Food & Drink, Music, Outdoors, etc...
  • Celebrities and Influencers - Academics, Activists, Actors, Artists, Athletes, Comedians, etc...
  • Organizations - Activism, Education, Sports Teams, Government, etc...
  • Entertainment - Book Publishers, Fan Accounts, Fictional Characters etc...
  • Miscellaneous - City and Local Accounts, Events & Awards, Meme and Humor Accounts, Motivational, etc...

We do periodically refresh and improve categorization, so the categorization may not always match exactly to what you see in the above specification.

What constitutes a brand?

Categorization can be a fuzzy and nuanced business. We do our best to provide as clear a distinction as possible, though it can be challenging at times. Our definition of brands is companies from which consumers can explicitly purchase goods and services from. Due to the varied uses of social media, there are plenty of examples which straddle the lines of being defined as a brand. We provide a couple examples below to help demonstrate and give you a good sense of how this works.

  1. Walmart (@walmart) - You can explicitly purchase goods and services from Walmart. Walmart is classified as a brand.
  2. Atlanta Falcons (@atlantafalcons) - There are a number of great arguments that the Falcons are a brand. Despite these arguments, to the average person they are professional sports team first and a business second. We classify them as "Top Social Media Accounts > Sports Teams".
  3. Google (@google) and Made by Google (@madebygoogle) - Several entity's have more than one account. Both of these accounts are classified in the Brands category. However, these accounts also show up in the Interests category under the Google Enthusiasts variable.

Don't hesitate to reach out to us if you have questions or feedback on our categorization. We're always trying to improve.


Back to top