Meta's "Phantom Ads" are Back and They're Worse Than Ever

Meta's Phantom ads are now promoting offers that do not exist any more — and your brand may be caught up in it. Also: What science says is the perfect influencer video. And more!

October 21, 2022

Good Lord, It's Happening Again.

The Facebook phantom ads are back, and now they're worse than ever.

You might recall our coverage of this last month. Meta was plucking uploaded creative from the ad accounts of brands, then creating a very basic ad out of it automatically, and running it as an ad on the platform.

They were terrible ads. An image, and the body copy reading "Shop" and then the brand name "Today."

The brands didn't pay for these ads, but they also didn't consent to them. And since it was, of course, an automated thing, sometimes these ads would be promoting seasonal products that the brand did not want to run. Or they'd run one brand's copy with a different brand's product photo.

The worst part of this was Meta's reaction to both our coverage and industry reaction.

When we first asked about what was happening, Meta's media relations told us it was a bug. When we pointed out that Meta engineers on Twitter were saying they were doing it on purpose, their media relations team stopped replying to our emails. When I mentioned to one of these Meta engineers on Twitter that running ads for brands without their permission wasn't a great idea, he blocked me.

Meta never did explain what was happening, formally, though apparently very high spenders were getting a little bit of info from their Meta reps. Eventually, they put up a help page on their site.

Well, those phantom ads appear to be back — and now, they're even worse.

One marketing consultant in Andrew Foxwell's Slack group reported this today:

Meta is running an ad for a client’s Shop with an image from a promotion that is no longer live. The image says people will get a free product with purchase. They won’t because that promo ran months ago.

Speculation in the group was that Meta's automation might be looking at phantom ads that performed well earlier, and just re-running them, without any consideration for — well, anything. Not seasonal promotions. Not inventory levels. Not creative alignment.

And brands were automatically opted in without any notification.

There is a way to opt-out, if you want, and it's in Commerce Manager:

  • In the left hand menu, click on Settings, and then click General.

  • Look for the box marked Promote products from your shop

  • Click the Edit button.

  • Click Turn off. 

  • Click Turn off again to confirm

  • Finally, add one more pin into your Zuckerberg voodoo doll

When "Banned Ads" Aren't

So while Meta is putting ads up without brands’ permission, a new report says they're also letting ads that should be prohibited into distribution. 

🔍 The Investigation 

Ahead of the U.S. midterm elections, Global Witness and the Cybersecurity for Democracy team at NYU investigated Facebook, TikTok, and Youtube's abilities to detect and remove election disinformation.

All ad content tested by the researchers contained outright false election information, such as:

  • False information about when and where to vote

  • False methods of voting (e.g. voting twice)

  • As well as delegitimized methods of voting such as voting by mail

🕵️ The Findings

The researchers used identical ads on all three platforms and submitted 20 ads in both English and Spanish. So, how did the social giants fare? 

  • TikTok fared the worst. The platform, which by the way does not allow political ads, approved 90% of the ads that contained outright false and misleading election misinformation.

  • Facebook was only partially effective in detecting and removing the problematic election ads, and approved a “significant number." 

  • Only YouTube succeeded both in detecting the ads and suspending the channel carrying them.

After passing the approval process, the researchers withdrew the ads, so users weren't exposed to misinformation-containing ads.

👀 The Response

In response, a Meta representative said the tests “were based on a very small sample of ads," and are not representative given the number of political ads Meta reviews daily. They also added that the company's "ads review process has several layers of analysis and detection, both before and after an ad goes live." Suggesting if it went live it would have been removed. 

TikTok's response, on the other hand, doesn’t account for much. According to a company spokesperson:

 [TikTok] is a place for authentic and entertaining content which is why we prohibit and remove election misinformation and paid political advertising from our platform.

Right... 

According to the researchers, the only ad that TikTok rejected claimed that voters needed to receive a Covid-19 vaccination to vote. Meanwhile, Facebook accepted that submission.

Images: Cybersecurity for Democracy

The Perfect Sponsored Videos Formula

Sponsored videos. They’ve become an important marketing tool as video-sharing platforms and the popularity of influencers have grown. But, to date, not much research has been done around how the design of sponsored videos affect viewer engagement. 

That research gap is part of the reason Lee Chen looked into it. She is a professor of marketing at Suffolk University in Boston. She and her colleagues recently published a paper in the Journal of the Academy of Marketing Science called “What drives digital engagement with sponsored videos? An investigation of video influencers’ authenticity management strategies.”

I spoke with her earlier today and asked her if it's better to have an influencer mention your brand at the beginning of their video, or the end.

LEE: Based on our findings, it's better to talk about brand information at a later stage in your video. While you can clearly disclose sponsorship, talking about the brand in the later part of your video is better for your engagement.

TOD: What about influencers' subjective endorsements, like sharing personal experiences or opinions about the sponsored product?

LEE: That's another interesting finding. There are two types of information you can provide: subjective information (talking about your own opinion, your own experience, your own thoughts about the brand), the other is objective (focus on the facts about the product.) 

In our study, we found that subjective endorsement, meaning that you talk about your own experience with the brand negatively impacts digital engagement.

TOD: Oh, interesting. That surprises me.

LEE: That's also surprised us a little bit. We believe it's because subjective endorsement is hard to verify. It's your own experience. For example, endorsing a snack and then telling you 'It's really delicious, you really should try it.' That's my subjective opinion. But it's hard to verify. As viewers, they just don't know whether it tastes good or not. 

But objective information is verifiable. For example, 'This snack is made from organic ingredients,' then I can easily verify the information. But subjective endorsement is not. 

Our full conversation covers all their findings, including:

  • How disclosure affects engagement

  • If there's a difference between influencer disclosure, or platform disclosure, like that "Sponsored Post" tag

  • Whether this effect is stronger on longer form influencer videos, like on YouTube, or on shorter ones like on TikTok

  • How engagement is affected when the influencer deliberately uses low production values

  • And a lot more.

That full conversation is coming tomorrow exclusively to the Premium Feed. You can sign up now and listen to the full episode, and more than 20 other interviews with marketing scientists that you will not hear on any other podcast. Plus: This daily feed will become ad-free, you'll get it earlier in the day, and a bunch more enhancements. You can learn more about it at https://todayindigital.com/premiumfeed or tap the link in the show notes.

Uber: Your Ad Can Fill the Awkward Silence

Buckle up – Uber ads are coming. 

The ridesharing company recently announced advertising options that will let marketers target riders based on their location at every point during their use of the app. 

🚕 Journey Ads

According to the company, the new ad-targeting product will allow brands to place ads based on riders' recent travel history and precise geographical location. For example, if a user books an Uber to a specific retailer, cinema, or airport, your brand can buy ads around that location.

Brands can also sponsor an entire trip, The Wall Street Journal reports. These "journey ads," which will be sold on a per-trip basis instead of the usual digital advertising pricing based on consumer impressions, will let brands show different ads to users at three points along their journey: 

  1. While waiting for a car

  2. While riding in the car

  3. Upon reaching the destination

The rider can also conduct transactions, such as clicking the ad to buy a product without leaving the Uber app. Separate pilot programs in the U.S. and India will also include ads on in-car tablets.

These ads will be available in select markets at launch.

🔒 Privacy Concerns

This, of course, is already raising privacy concerns. 

An Uber spokeswoman said the company never shares individual users' data with advertisers. She added that the data it does share is aggregated information or data related to ad-campaign performance. Users can also opt out of targeted ads on the Uber app at any time.

According to The Wall Street Journal, Uber's advertising policy forbids targeting users by factors such as race, religion or sexual orientation, and it also prohibits targeting users by certain kinds of destinations, such as government buildings, hospitals and reproductive health centers.

Snap’s Woes

Snap reported mixed earnings yesterday, with users increasing but ad revenues slowing.

👍 The Good

First, the platform added 16 million more daily active users for the quarter, taking it to over 360 million. 

The app has seen significant growth in India, which as Social Media Today points out is positive for its long-term potential, but doesn't significantly boost its revenue at the moment, like adding U.S. users would.

Moreover, the company also reported that overall viewing time in the app continues to increase, driven primarily by Discover and Spotlight.

On Spotlight specifically, Snap indicated that total time spent watching content on its TikTok clone has increased by 55% year-over-year.

👎 The Bad

Now for the bad news. 

Quoting the company:

Total time spent watching content in the United States decreased 5% year-over-year as the diminished depth of engagement with Friend Stories was not fully offset by the growth in viewership and growth in time spent with Discover and Spotlight in the US.

Considering Snap's revenue statistics, this is especially relevant. Snap's revenue for the quarter increased by 6% year-over-year to $1.13 billion, but its income is heavily dependent on the U.S. market. It was also the slowest pace of growth since the company went public.

Snap pointed its finger at advertisers who continue to cut marketing budgets because of macroeconomic headwinds. It also blamed increased competition and everybody's favourite kicking-horse — Apple — with its previous privacy changes making it difficult for apps to target advertising and measure results.

Images: Snap

Musk’s Final Boss... Joe Biden?!

It looks like Elon Musk's final boss level in the Twitter deal could be Joe Biden. 

Bloomberg reports that U.S. government officials are considering conducting a national security review of Musk's Twitter deal, which could lead to Biden blocking the deal.

Sources say Musk's plans to purchase Twitter for $44 billion with the help of foreign investors have concerned Biden administration officials. Now, officials are reportedly considering which tools they could use to review Musk's ventures, including the use of the Committee on Foreign Investment, which reviews foreign takeovers of American businesses.

A review of the Twitter acquisition by this committee could result in a recommendation to President Biden to nix it… Just as Elon Musk wanted. 

A Twitter user comments on the Bloomberg article tweeting "it would be hysterical if the government stopped Elon from over paying for Twitter." 

To which Musk replied with the "100" emoji and a rolling-on-the-floor-laughing emoji.

🗣️Twitter Squashes Layoff Rumours

In other news, Twitter is confronting reports that Elon Musk is planning to gut the company's workforce if the deal goes through. 

The Washington Post reported yesterday that his plan for Twitter involves slashing its staff by 75% in a matter of months. 

But now, Twitter has told staff that there haven’t been plans for companywide layoffs since it signed a deal to be acquired by the billionaire, according to an internal company memo seen by Bloomberg.

The memo warned workers to expect “tons of public rumours and speculation” as the closing of the deal nears and that company does not have "any confirmation of the buyer’s plans following close and recommends not following rumours or leaked documents but rather waiting for facts.”

In Brief

LINKEDIN has wiped out 50% of accounts that claim to be Apple employees in a new effort to kick bots off the platform. The professional network has seen an uptick in fraudulent activity with fake accounts claiming affiliation with organizations they were not affiliated with, along with altered photos and profile descriptions lifted from legitimate profiles.

CONSUMERS plan to shop more in stores this holiday season, with nearly two-thirds planning to shop IRL. This is followed by 55% who plan to shop with an online-only retailer, while half expect to order online from a brick-and-mortar store. Among the reasons customers prefer to shop in stores is to view products before buying them and to avoid shipping costs and delays.

INSTAGRAM launched new safety tools yesterday to enhance filters on DMs, block other accounts from someone you've blocked, and it will even start reminding you to be nicer in comments and DMs. Its "hidden words" DM filter tool now expands to Story replies and claims to catch intentional spelling mistakes. These DMs will go directly to the Hidden requests folder.

AI ART: You've probably heard about DALL-E, the AI that can generate artwork from a text prompt. There is a public-facing version of it up now that is free to use, though it only generates Hallowe'en images. Still, it's fun to play with, and you might end up making something that you could use on your brand's site or email list. It's at artbot.ai if you want to check it out.

Reply

or to participate.