Oddly specific ad profiles

Advertising funds a big chunk of the web, but for advertisers to continue to spend, their placements have to deliver results. So companies collect data about people’s online activity and create profiles based on the behavior. For The Markup, Jon Keegan and Joel Eastwood, dig in to the specificity of these profiles.

Profiles get stuck in segments or groups, and advertisers can choose which segment to put ads in front of. The above are finance-based segments. I’ve always dreamed of being a “Silver Sophisticate” myself.

You can download the data the project is based on here.

Tags: , ,

Election ad topics

Midterm election day is just about here in the U.S., so the political ads are running. Harry Stevens and Colby Itkowitz, for The Washington Post, show the spending breakdown by political party and topic. Bigger squares mean more spending, and more blue or more red mean more Democrat or Republican, respectively, share of the spending.

The chart reminds of the Shan Carter classic from 2012, which visualized word usage at the National Convention. Same split and sort, but with circles.

Tags: , , ,

Trump’s criminal justice ad spending on Facebook

The Marshall Project contrasted ad spending on Facebook by Trump’s campaign against Joe Biden’s:

Our analysis found that of the $82 million Trump’s reelection campaign has spent on Facebook ads this year, $6.6 million paid for ads about crime and policing—a top focus of his Facebook campaign. Almost all of it came since George Floyd was killed by police in Minneapolis in May. More than one-third of those ad buys were aimed at key battleground states and many sought to persuade specific undecided voters, and married women in particular. The Biden campaign? It didn’t spend a cent on criminal justice ads on Facebook until late August, choosing instead to focus on the COVID-19 pandemic and economic recovery. Yet Biden had, during the Democratic primaries, articulated a more progressive criminal justice platform than any of his party’s recent nominees.

Tags: , , ,

Maps of the issues mentioned most in election advertising

As the midterm elections loom, the ads focusing on key issues are running in full force. Using data from Nielsen, Bloomberg mapped the issues talked about across the country.

Bloomberg News analyzed more than 3 million election ads for 2018 congressional and gubernatorial races to get a sense of the most commonly discussed issue in 210 local television markets, as defined by the Nielsen Company. Across the U.S., 16 different topics are mentioned more than anything else during midterm TV ads.

The map above shows the most common per Nielsen market, but read the full article for the national breakdowns of the major issues.

Health care has been huge in my area. For the past few weeks, every YouTube video I watch is preceded by an ad, and my mailbox keeps getting filled with ads for and against a certain proposition, often on the same day.

Tags: , ,

Facebook still allowed race exclusion for housing advertisers

Last year, ProPublica revealed that Facebook allowed housing advertisers to exclude races in their campaigns. Facebook said they would address the issue. ProPublica returned to the topic. Facebook didn’t do a very good job.

All of these groups are protected under the federal Fair Housing Act, which makes it illegal to publish any advertisement “with respect to the sale or rental of a dwelling that indicates any preference, limitation, or discrimination based on race, color, religion, sex, handicap, familial status, or national origin.” Violators can face tens of thousands of dollars in fines.

Every single ad was approved within minutes.

Ugh.

Tags: , ,

Infinite Twitter ad campaign, based on data profiles

As you probably know, Twitter (and all social media) collects data about you and infers your likes, dislikes, wants, dreams, hopes, etc. Sam Lavigne set up a scraper to find out all the user segments, ranging from “buyers of cheese” to “households with people who have recently moved into a new home.” It can get pretty detailed. Lavigne then used this data to automatically generate an infinite ad campaign, on what else, Twitter.

Using this list I wrote a program, “The Infinite Campaign”, that automatically generates and posts an infinite series of video ad campaigns. The script randomly selects two behavior categories and one interest category from the ad creation page. It rephrases the descriptions of the categories, putting the statements in the second person. The Infinite Campaign then overlays those statements on top of automatically selected stock footage. Finally, it logs me in to Twitter, uploads the video, and auto-generates a new ad campaign, targeting the same behavior and interest categories used to generate the video.

The data is available for download, which includes the size of user segment and the data brokers involved.

By the way, you can opt out of some of the tracking in the privacy section of your Twitter settings. Obviously that doesn’t stop others from tracking you, but at least it’s something.

Tags: , ,

Tracking online ads

Floodwatch signature mode

We browse online, we see ads, and we buy stuff. The better-targeted the ads are, the more likely that we buy stuff. So of course advertisers continue on ways to guess who you are and what you might want to increase the chances that you click and spend. Floodwatch, a Chrome extension by the Office for Creative Research and Ashkhan Soltani, lets you turn it around ever so slightly so that you can track what the advertisers serve you.

Leave the extension on, and see banners advertisers served you, along with metadata such as size, time, and the site an ad was on. You can also compare your banner set to what other Floodwatch users see and filter by demographic.

That second view, shown above, is in line with the overarching goal of Floodwatch. Anonymously share your data to help researchers decipher the algorithms behind the advertising black box.

Extension installed and enabled.

Video below for more on the project:

See also the writeup by Jer Thorp from OCR. He showed his banner profile to ten anonymous people on Amazon Mechanical Turk and asked them to guess the type of person he was — lonely but likely living an exciting life and Jewish.

Tags: ,