Digital Political Ads Are Poorly Targeted - And Increasingly Important

Social media companies and Google are not required to explain how and why online political ads are targeted because they claim the algorithms are proprietary competitive assets.

Which means campaigns may not be getting the results they want for the money they're paying and consumers have no influence over how and why they are targeted. JL

Sara Morrison reports in Re/code:

If you want to find out how or why you’re being targeted, good luck. Most answers are hidden beneath layers of ad technology and data collection that tech companies aren’t legally required to disclose. And unlike TV, print, and mailed political ads, which are regulated, online political ads are unregulated. That means it’s up Facebook and Google to make their own rules. It is impossible to know why you’re seeing a certain ad on your own timeline. Even the campaigns themselves have to trust Facebook that the audience they’re asking the social media company to reach is the one that sees their ads.

As November 3 gets closer, election ads are almost impossible to avoid. And this year, we’re seeing more of them than ever online, peppered throughout our Facebook timelines, Google searches, and in Hulu videos. Some of them may not apply to you, but all of them — like every other online ad — are targeting you.

If you want to find out how or why you’re being targeted, good luck. Most answers are hidden beneath layers of ad technology and data collection that tech companies aren’t legally required to disclose. And unlike TV, print, and mailed political ads, which can’t target you nearly as well and are regulated by Federal Election (FEC) and the Federal Communications (FCC) Commissions, the online political ad world is largely unregulated. That means it’s up to companies like Facebook and Google to make their own rules governing the deluge of political ads we’re all seeing, and there’s no guarantee they’ll make decisions that protect our democracy.

Political ads are firmly entrenched in the digital world

It’s estimated that nearly $7 billion dollars will be spent on ads over the 2020 election cycle overall. A lot of that money is going to television ads, as is the case every year. But an increasingly large portion of it is paying for digital ads. Trump’s focus on digital ads during the 2016 election was a watershed moment for digital political ads, though it wasn’t the first time a presidential candidate made use of them. Barack Obama’s 2008 run was widely praised for its use of digital ads and microtargeting. It even put ads in video games. But the Obama campaign only spent about $8 million on online ads, and less than $500,000 of that went to Facebook. A lot has changed since then.

The Trump campaign spent tens of millions of dollars on millions of Facebook ads in 2016, sometimes running tens of thousands of ad variations a day to microtarget potential voters with messages tailored to their interests. The effectiveness of Trump’s digital operation is up for debate; either way, 2020 campaigns are devoting a lot more time and money to their digital ad strategies than they did in 2016.

Madeline Kriger, a director at a Democratic Super PAC called Priorities USA, told Recode that in 2016, her PAC spent the vast majority of its ad money on television. She said that was almost “universally true among Democratic groups at that time.” But it isn’t anymore. Priorities spent the last three years practicing and refining its digital strategy for 2020, including bringing much of its ad operations in-house rather than rely on an outside group to do the buying and placement.

“We have now built out an entire internal ad-buying team,” Kriger said.

On the other end of the spectrum are the smaller, less-moneyed campaigns that benefit from targeted digital ads that give them access to potential voters that may be receptive to specific messages, and often cost less than other mediums. A campaign for a state senate seat, for example, can target Facebook ads stressing its candidate’s concern for the elderly to older constituents — which is exactly what Martha Marx, a Democrat running in Connecticut Senate District 20, is doing.

“I target each ad to different groups depending on the ad or need at the time, while simultaneously running some ads district-wide,” Jason Ortiz, Marx’s campaign manager, told Recode. “I tend to oversample folks who meet the same demographic as the messenger in the ad.”

Facebook and YouTube ads are a great way to reach voters, Ortiz said, and at a fraction of the cost of a television ad.

The campaign has spent about $15,000 on digital ads during this campaign cycle, Ortiz said. Marx’s opponent, incumbent Republican Paul Formica, has only spent a few thousand dollars so far according to Facebook and Google records. In what should be a close race — two years ago, Marx lost to Formica by the slim margin of about 1,500 votes out of the 40,000 cast — spending thousands more than her opponent on campaign ads could give her the edge.

But some campaigns are expanding beyond Facebook and Google. Now, they’re frequently running ads on streaming platforms like Hulu and Pandora, and on ad networks like Verizon Media. They’re finding alternate ways in, like how Democratic presidential candidate Mike Bloomberg paid Instagram influencers to spread the word about his campaign on their accounts, or Democratic presidential nominee Joe Biden’s campaign’s Animal Crossing yard signs (Biden’s campaign told Recode that it didn’t pay Nintendo to be able to include the signs in the game, so these are not an ad so much as they are the digital equivalent of campaign yard signs).

“It’s a real mess”: Digital platforms get to make their own rules

Without online-specific regulations, platforms are making their own rules, which can vary widely, are voluntary, and can be changed at any time.

Some platforms have chosen not to allow any political ads at all, including Twitter, LinkedIn, TikTok, and Spotify. Others have clamped down on who can buy political ads and how they have to present them. Reddit has a library of candidate or issue-based ads, manually reviews all of them, forbids deceptive ads, requires certain disclosures, and only allows ads from candidates for federal office. Snapchat’s political ads policy forbids misleading or deceptive ads, and the company also provides a list of political ads, including the audiences they target, in a downloadable spreadsheet.

As for the big guys: Google has placed several restrictions on ad targeting — only location, age, and gender can be used — and has a political ad transparency section, which includes a library of paid ads. Facebook, which recently announced a moratorium on new political ads in the week before the election, also has a transparency section that includes a political ad library, but it lets political ads target users the same way most of its other ad categories can. The company has also said it won’t forbid politicians from lying in ads, with CEO Mark Zuckerberg telling Rep. Alexandria Ocasio-Cortez, “in a democracy, I believe that people should be able to see for themselves what politicians that they may or may not vote for are saying and judge their character for themselves.”

Facebook does limit ad targeting on ads for housing, employment, and credit, but that’s only because of the company’s 2019 settlement with civil rights groups over how the platform could be used to enable illegally discriminatory ads — for instance, by targeting job ads to young white men only. (The 2016 Trump campaign made use of these options by targeting anti-Clinton ads to young women and Black people that contained allegations of sexual assault against Clinton’s husband and her 1996 “superpredators” comment, respectively.)

Facebook has since removed race and ethnicity as an advertising target, but there are ways to get around this. Facebook itself says it “encourag[es] advertisers to use other targeting options such as language or culture to reach people that are interested in multicultural content.”

Facebook does require any entity that places an ad regarding social issues, elections, or politics to submit to various identity verification measures, and to put certain disclosures on those ads. Its ad library says who is paying for which ads, how much they’re paying for them, and the age, gender, and location of users who saw the ads. But it does not — and will not — provide specifics on advertisers’ targeting requests.

“I do give [Google and Facebook] credit for spending the last couple of years building those out to be more transparent than what they started out as, and actually helpful for understanding what various political actors are doing,” Kriger said. But, she points out, there are many other places for digital ads now that aren’t transparent at all because they’re not legally required to be.

“You really don’t know what a campaign or an organization like ours, frankly, is doing,” she said.

Those places include ad-supported streaming video services like Hulu, Sling, Roku, and Tubi, which have replaced broadcast television for many Americans and promote their audience-targeting abilities to potential advertisers. Those services are far less transparent than their broadcast peers and have fewer disclosure requirements, such as keeping public archives that say who paid for an ad, how much they paid, and where and when the ad appeared. A recent report from Mozilla shows that they haven’t done much on their own to promote that transparency.

“It’s a real mess, with a lot of spending not being disclosed, and it really shows the need for a uniform national standard,” Adav Noti, senior director of trial litigation and chief of staff at the Campaign Legal Center, told Recode.

What digital political ad rules could look like

It wasn’t until the 2016 election cycle that the federal government started to take a serious look at political ad disclosure rules and whether they needed to be updated for the digital world, Noti said. But almost four years and another presidential election cycle later, the federal efforts have gone nowhere. One attempt, the bipartisan Honest Ads Act, would have extended television and radio ad rules to cover the internet, required large digital platforms to keep a public file of political ads, and forbade foreign actors from purchasing digital political ads. That bill was initially proposed in 2017 and reintroduced with bipartisan and bicameral support in 2019. It passed the House but Sen. Mitch McConnell has so far refused to put it up for a vote in the Senate.

The Federal Election Commission could also update or clarify its rules to better reflect the online era, but it needs at least four of its six members to agree on what those updated rules would be, and it currently has only three members. FEC members are appointed by the president.

In the absence of federal rules, some states have stepped up with their own laws. Washington state passed its own transparency legislation, prompting Facebook and Google to forbid political ads in the state rather than follow the requirements. Showing how difficult it can be to ban something on the internet in a particular region, political ads have shown up in those states anyway. Maryland’s attempt to create a similar law was struck down by a federal appeals court for violating the First Amendment.

While digital political ads are required to follow some political ad disclaimer rules, that’s not enough for advocates who believe the lack of regulation makes it impossible to monitor them for potential abuses. An ad archive requirement and targeting disclosures would add necessary transparency and update the rules for the digital age, experts say.

“If you have an archive, then law enforcement can look at the archive to look for leads on potential law-breaking, or watchdog groups like ours can look at the archive and try to ferret out sketchy behavior,” Noti said. “Obviously, political opponents can look at the archive and engage in counter-speech if there are lies or misinformation being spread.”

Noti also believes that transparency should include information about how the ad was targeted.

“You want to know what criteria were used to target that ad,” Noti said. “That’s a much more relevant piece of information for digital ads.”

Why am I seeing this political ad? You’ll probably never know.

But it’s a relevant piece of information that Facebook refuses to provide, which means that the ways campaigns are targeting potential voters on Facebook and its properties (including Instagram and the audience network, i.e. ads that show up on outside websites) remain unknown to the people who are seeing them — not to mention opposing campaigns and watchdog groups. Who sees those ads are determined by Facebook’s algorithms — also secret — which use the immense amount of data the platform harvests from its users’ activities both on Facebook and off (including on many campaign websites). Alternately, you might be included in what’s called a “lookalike audience,” where an advertiser sends Facebook a list of users it thinks will be receptive to its ad, and Facebook sends the ad to a much larger audience that mirrors some of the source group’s qualities in the hope that they will be similarly responsive.

You can stop some of Facebook’s targeting by turning personalized ads off on Facebook (Google also gives you this option, as do most other platforms) — but you can’t stop all of it.

Facebook can still figure out your approximate location and target location-based ads to you based on your IP address, for instance, so you’ll still see ads for local candidates. And there are other ways campaigns can target you on Facebook (and other platforms, like Snapchat) without even using their targeting services. They may create audience lists, or, as Facebook calls them, “custom audiences,” which may include personally identifiable information like your email address, name, or other data points. (Google doesn’t allow audience lists to be used for political ads.) Those are then matched to user profiles, and the ads are sent to them. So even for all of Snapchat’s transparency about political ad targeting requests, it can’t tell you any targeting that comes from audience lists.

Those lists may come from information you knowingly and willingly supplied, like signing up for a campaign’s mailing list, but they might also come from information purchased through data brokers, which have their myriad ways of acquiring that information, often without your knowledge. Or they might come from public voter files, which can have a lot of information about you, including your age and address. Campaigns have been relying on these databases for decades — long before Facebook ever existed — but the internet has made them that much more useful.

All this makes it almost impossible to truly know why you’re seeing a certain ad on your own timeline. Even the campaigns themselves have to trust Facebook that the audience they’re asking the social media company to reach is the one that sees their ads.

“It’s very much a black box, honestly,” Kriger, of Priorities USA, said.

For her part, Kriger wants to see more regulation of digital ads, rather than the current patchwork of self-imposed, inconsistent, and ever-changing policies her PAC has to navigate now to get its digital ads in front of voters. Priorities USA had one ad accepted by Facebook that was rejected by Google, Hulu, and Verizon, and another ad Facebook erroneously rejected. On the other side of the aisle, misleading ads about mail-in voting from a mysterious group associated with Republican causes were pulled from Facebook but allowed to stay on Google.

Even if digital paid political ad legislation that made every side happy were passed, there’s a whole other side that it won’t address: all the things that politicians, campaigns, and committees can say on social media that spread to as many people as possible because that’s what those platforms are designed to do. Trump can’t buy an ad on Twitter, but he can and does tweet ads to his 86 million followers all day long, and he doesn’t have to pay a penny to do it.