The weirdly particular filtering campaigns are used to focus on you to micro-targets


The news: The NYU Ad Observatory released new data this week on the inputs the Trump and Biden campaigns are using to target audiences for ads on Facebook. It's a jumble of broad and specific traits, ranging from extremely broad ("users between 18 and 65 years old") to specific traits (people with an "interest in Lin-Manuel Miranda"). Campaigns use these filters – usually several in each ad – to target ads to segments of Facebook users in order to persuade, mobilize, or raise funds. The data shows that both campaigns have invested heavily in personality profiling using Facebook, similar to what Cambridge Analytica claimed in 2016. They also show how personalized targeting can be: campaigns can upload lists of certain individual profiles that they want to target. The study shows that this is a very common practice.

Here's how targeted ads work: Campaigns create public relations strategies for voters by using models that process data and make predictions about the likelihood of people voting. From this they identify from which of these segments they want to collect money, convince them or participate in the surveys. Facebook offers advertisers a number of ways to target these users, including basic demographic filters, a list of user interests, or the option to upload a list of profiles. (Facebook creates the list of topics users might be interested in based on their friends and their online behavior.) Campaigns use personality profiles to match their segments to Facebook interests.

However, when campaigns upload lists of specific users, it is much less clear how they determined who to target and where the profile names came from. Campaigns often buy lists of profile names from third parties or create the lists themselves. However, it is extremely difficult to see how a campaign matched a voter against a Facebook profile.

The data: The data is not comprehensive or representative as it comes from approximately 6,500 volunteers who downloaded the Ad Observatory plugin. Facebook does not publish this data, so voluntarily sharing it is the only window in this process. This means that it is difficult to make a fair comparison between the campaigns or to take a comprehensive look at their activities. Working with the Ad Observatory team, we were able to pull out some examples of filters and provide the ad pairing for the audience included in this story. You can examine the rest of the data at the bottom of this dashboard.

How to interpret it: The NYU researchers say there are some lessons to be learned. First, it is clear that campaigns continue to experiment and invest in targeted advertising campaigns on Facebook. The researchers also said that ads created with custom lists were more likely to be used for compelling news. It is unclear why this is exactly, but there is a lucrative industry out there that finds voters and sends them messages that could potentially be persuaded.

Most of the ads generated with the specific interest filters were for donation purposes, but not all. Fundraising ads are aimed at grassroots supporters. As a result, campaigns can have more complex models (and better data) when it comes to the interests and personalities of their own supporters.


Steven Gregory