Inside the ‘made for advertising’ crackdown—and why it isn’t working

Trading in “made for advertising” platforms remains rampant despite months of solemn vows by ad tech platforms and agencies to eliminate or curtail it, according to a report published today by research firm Adalytics. The Association of National Advertisers set off a wave of scrutiny and condemnations regarding MFAs in June with a first look at its programmatic marketplace study that found 21% of impressions and 15% of spending in its sample was on the low-cost, low-quality sites. MFAs do well on viewability and cost-per-thousand (CPM) metrics but fare poorly on generating sales or brand lift, according to agency, tech and marketer executives. They’re characterized by high volumes of paid traffic, click-bait content and pages heavily stacked with ads. And numerous ad tech players and agencies appeared to entirely forswear MFAs or announce steps to reduce trading in them in the wake of the ANA report. Read more: A brand guide to MFAs Ad Age A-List and Creativity Gala Buy your ticket to the in-person event Click here Yet in January, seven months later, Adalytics found 16 of the 17 named ANA members that participated in the study still had ads showing up on MFA sites. Ad tech players and agencies that had criticized MFAs were still largely involved in buying them, according to the report. That includes 12 of 15 supply-side platforms invited to a “Meet the SSPs Virtual Supplier Showcase” hosted by the ANA on March 7, according to the firm. The exceptions among that group, which didn’t show up transacting MFA inventory in the Adalytics study, were Kargo, InMobi and Freewheel. One SSP in attendance (see below) was found by Adalytics to have transacted ads for 11,671 brands on the Adalytics sample of 21 MFA websites in January. The report found all seven of the leading agency holding companies, including ones that have been most vocal about efforts to avoid MFAs, were involved in at least some MFA transactions on behalf of brands. Likewise, numerous SSPs and DSPs were linked to MFA deals through source code found by Adalytics in its research, including from SSPs that appeared to issue statements categorically excluding MFA inventory from their offerings. However, Adalytics found source code or other data linking most of the agency and ad tech world to MFA deals in January. No party contacted by Ad Age denied instances of deals cited in the firm’s draft report, though some indicated they represent very small percentages of the business they handle. The Adalytics data don’t indicate whether the deals were authorized by clients explicitly or implicitly. They also don’t show whether ads were placed on MFA sites intentionally or because of glitches or new workarounds by MFA sites to avoid detection. Advertisers contacted by Ad Age about the study generally said their policies call for avoiding MFAs, and they rely on agency and tech partners to follow through. Agency and tech platform representatives said they exclude MFAs from buys wherever possible when clients request. Regardless, breaking up with MFAs appears hard, though the report also found a handful of tech companies where MFA activity was never detected. Those include the Ozone Project, Kargo and TrustX SSPs and The Trade Desk DSP. “It’s good, easy money for everybody in the industry, and people are having trouble giving it up,” said one ad tech executive about MFAs. But others noted problems that range from marketers demanding cost-per-thousand targets that only can be delivered through MFAs to rapidly shifting tactics by the MFAs themselves to elude detection or shift sites in and out of MFA definitions as often as daily. “It was a painful realization of where we are as an advertising industry,” said one marketer executive, who had read the draft report and spoke on the condition of anonymity. The marketer noted the apparent “inability of the tech platforms to do what we need to do, and probably a wake-up call for advertisers to hold them to account.” Also read: Advertisers waste 23% of programmatic ad dollars Amazon effect Even marketers that seem to largely avoid MFAs, such as Procter & Gamble Co., appear to be buying them anyway through retail media, specifically Amazon, according to Adalytics. The study found ads transacted through Amazon (outside of Amazon.com websites but purchased through the Amazon demand-side platform), widely appear on MFA sites, including some from P&G, which declined to comment. In all, Adalytics reported it found ads purchased through the Amazon DSP on MFA sites for 2,601 brands in January. After deadline for this story, Amazon issued the following statement: “We maintain a high bar on the supply that is available for our advertisers and proactively block made-for-advertising inventory available through a combination of manual and automated processes. We also enable advertisers to decide where their ads appear through a mix of proprietary and third-party controls. We will continue to build the Amazon DSP as a service that provides advertisers with the relevant reach and control that delivers the marketing results they want.” One marketer, who had read a draft of the Adalytics report and spoke on the condition of anonymity, said retail media has grown so fast that platforms, including Amazon, can’t deliver inventory to meet the demand without resorting to low-quality MFA inventory. Another factor behind the problem, the marketer said, is that money for retail media largely comes from trade promotion budgets, not brand marketing budgets, and is effectively a “slotting fee” to get or maintain distribution and improve position on the digital shelf. As such, the marketer said, it’s often not subjected to the same performance expectations as brand spending. In other words, brands often spend on Amazon or other retail platforms because they feel they have to, not because they want to, the marketer said. But the marketer said it’s time for brand advertisers to get more assertive, even with retailers that hold considerable power. “We need to step up as a community and have more of a voice,” the marketer said. “We’ve struggled with [Amazon] to even get them to understand what MFA is and how it can be accessed,” said another marketer. “So from that perspective, they are quite far behind in regard to anything they’re doing that sits outside Amazon.com.” But Megan Pagliuca, chief activation officer of Omnicom Media Group, which was also cited in the Adalytics report for involvement with MFA deals, said it’s possible to trade through the Amazon DSP while avoiding MFAs. “We have the same Amazon inventory curation capabilities that we do with The Trade Desk or any other platform,” Pagliuca said. “So I think maybe that’s a problem for entities that do not have advanced inventory curation.” As for the other instances noted by Adalytics in its draft report, she said they accounted for well under 1% of client spending, which she saw as an indication that OMG’s systems are working. Also read: A guide to retail media networks Trade Desk and Walmart Connect avoid MFAs Amazon was the primary retailer singled out in this report, which also notes that The Trade Desk tags didn’t show up in source code for any of the MFA ads tracked in the research. That includes for Procter & Gamble Co., which uses TTD for most of its digital buys. The report warned that this doesn’t mean TTD has no exposure to MFA trading, just that it didn’t show up in the sample (which did turn up involvement by dozens of the other trading platforms). TTD’s apparent relative success in avoiding MFAs in the study is likewise true for the Walmart Connect retail media business and its sibling Sam’s Club Member Access Platform, which use TTD for trading in the U.S. Ads transacted through Walmart retail media carry TTD tags in their source code, in addition to Walmart tags, according to Krzysztof Franaszek, CEO of Adalytics. Those tags didn’t show up in any MFA ad placements captured as part of the study, and a person familiar with the matter confirmed that any exclusions TTD uses to avoid MFAs should similarly apply to Walmart Connect. Another person familiar with the matter said TTD instituted controls against MFAs more than two years ago, when the issue was first surfaced broadly by the monitoring firm DeepSee.io, and uses another firm active in the MFA detection, Jounce Media, to screen out MFA sites. The Trade Desk declined to comment on exactly how it weeds out MFAs, but in a statement, Will Doherty, VP of inventory development, said, “We take great pride in our focus here, which is to ensure our platform provides access to the best possible inventory.” While Walmart Connect customers trade through TTD, Walmart doesn’t necessarily do so on its own account. Adalytics found cases of Walmart brand ads on an MFA site transacted through Publicis Media and DSPs other than TTD or the Walmart Connect platform. Walmart and Publicis declined to comment. Buying through TTD isn’t cheap. Some agency and marketer executives have complained privately about the cost, particularly on transactions that aren’t complex, and about the relative cost of trading through Walmart Connect vs. other retail media platforms. TTD executives have said the company’s “take rate” remains fairly steady at around 20%, which means its $1.9 billion in revenue last year represented its take on about $9.7 billion in digital and connected TV media spending. Related: Walmart’s retail media network names creative agency partners Massive frequency on some sites The report found visitors to MFA sites were exposed to, at times, thousands of ad impressions during a single page visit, eluding frequency caps. One of the high marks in the study came for a visitor shown 835 Kroger Co. ads in a single page view session at a cost of $5.49. Though it was only one visitor, that translates to a $5,491 effective CPM, even if each “impression” cost less than a penny. Kroger executives declined to comment. That effective CPM, for the individual, is well above the priciest inventory in the media world, including the roughly $70 CPM for a Super Bowl ad and $50 for Netflix advertising. Check out the latest Ad Age events and award programs A person familiar with the matter said instances cited in the report were not Kroger Precision Marketing retail media transactions, and that KPM has a self-service model through which brands buy the inventory of their choice using their own safety and verification standards. H&R Block had the highest single incidence of ad impressions for any brand shown on an MFA site in the report, with 2,117 impressions served to one visitor on one page. “We work with reputable and trusted partners in the marketing ecosystem and have implemented frequency caps across every single campaign and line item,” an H&R Block spokesperson said in a statement. “Our agency partners are investigating the error that appears to have occurred with one of our DSP partners and are making necessary adjustments.” In a statement, Sophia Zhang, managing director of H&R Block’s media agency, Gale, said: “We have an approach at Gale that enforces daily and weekly frequency caps across every campaign,” adding that the agency works with Integral Ad Science (IAS), Google, Amazon and The Trade Desk to enforce those rules. “We have isolated this issue to one DSP, and they are currently investigating the root cause of the buys in question,” Zhang said. “MFA is an acknowledged industry-wide challenge due to the way in which these bad actors are constantly shifting their approach and bypassing DSP policies and enforcement systems. Gale continues to work with our verification partner, IAS, to implement MFA blocking across all advertisers under their upcoming product release.” IAS last week announced an MFA exclusion product that remains in beta test. While MFA sites do not constitute fraud per se, which IAS and others offer products to block through other parts of the verification system, they present other problems, such as ridiculous levels of ad intensity and frequency. The Adalytics report cites unidentified ad tech executives describing how frequency capping is impossible with MFA sites, essentially because they overwhelm real-time bidding systems by generating thousands of simultaneous bid requests through multiple platforms for the same pages, causing deals to be executed before safeguards can work. Marketers and agencies still say they don’t want MFAs If there are marketers who actually want their brands on MFA sites, certainly none of them raised their hands to volunteer that when asked by Ad Age. “Placements on made for advertising sites do not align with our media strategy,” according to a statement from General Motors, in response to a request for comment on an instance reported by Adalytics of four Cadillac ads showing up on a page of an MFA site via Dentsu’s Carat and various tech platforms. “We continuously conduct media audits and optimize our programmatic placements to ensure we are delivering ads on sites that meet our standards.” GM Chief Media Officer Shenan Reed, who headed media for L’Oréal USA and chaired the ANA media committee when its report dealing with MFAs was issued last summer, declined to comment but confirmed through a GM spokeswoman that her views on MFAs align with ANA recommendations. Reed joined GM earlier this year. Dentsu declined to comment. State Farm, which had five ads appear simultaneously on the page of an MFA site cited in the Adalytics report, was one of the named participants in the ANA programmatic study released last year. A statement from the insurer read: “State Farm does not advertise on MFAs. Our independent contractor agents are expected to follow brand and marketing guidelines.” Omnicom Media Group, which handles State Farm as a client and has been vocal about efforts to curtail MFA placements, indicated that when cases of clients mentioned in the report were reviewed, MFA placements accounted for less than 1% of client spending. In one case, it appears a site identified as MFA by Adalytics was not classified that way by Omnicom, which uses data from Jounce and site quality rankings by IAS among its screening tools, Pagliuca said. “This is an area we feel like is a core strength,” Pagliuca said. Instances cited in the report, based on OMG’s review, amounted to only 19 impressions in one case and less than $7 in another. It’s common, despite all efforts, for some MFA placements to slip through, but OMG found none where there were substantial amounts, she said. WPP’s GroupM, which declined to comment, has been arguably the most vocal among agencies in the wake of the ANA report, issuing a statement in August about a new relationship with Jounce to help exclude MFAs. Nonetheless, the Adalytics report includes an example of Ford ads served on an MFA site via GroupM’s Xaxis DSP. GroupM declined to comment and Ford did not comment by deadline. Despite best efforts, identifying and excluding MFAs can be difficult, with definitions of some sites changing daily based on adding new subdomains and buying more paid traffic, said one executive familiar with the matter. In some cases, clients may have different expectations, or exclusion and inclusion lists, than the agency, the executive added. Other agencies cited in the report as having handled MFA transactions for clients also noted that they’re trying to exclude MFAs, but that the efforts may not always work. “While we continue to recommend all clients use Havas’ Meaningful Marketplace (MMP) solution, which prioritizes inventory that has been human vetted to optimize the user experience and ensure high quality supply, should clients choose to use a mix of our MMP and open marketplace, we are transparent about the inherent risks,” said Greg James, CEO of Havas Media Network North America. No technology solution is failsafe and the lack of standard definitions and parameters for measurement are among other hurdles, he said. Interpublic has rolled out a new process to reduce the prevalence of MFA sites in programmatic buying, including removing high carbon emitting websites from available inventory, and curtailing MFA placements as a result, said a spokesman. That includes using Jounce, Pixalate and its own artificial intelligence to screen out MFAs. Catch up on the latest agency news here SSPs on why their efforts often don’t work SSPs were among the most vocal parties in the industry about eliminating MFA inventory in the wake of last year’s ANA report. But those efforts clearly haven’t always succeeded, according to Adalytics. Magnite, which in July termed the ANA report indicative of a “troubling trend,” was found by Adalytics in January to have the largest number of MFA transactions among SSPs, involving 11,671 brands on the 21 websites tracked. “For more than six months, Magnite has blocked inventory categorized as MFAs by Jounce Media and Similarweb from all Magnite-sourced demand,” according to a company statement. “In open auction, however, we defer to individual DSPs. Some provide us with extensive MFA blocklists, while others are more tailored in their approach. Whatever their preference, Magnite transparently signals every URL to our demand partners, consistently enforces their blocklists, and uses a combination of people and technology to proactively filter anything that’s not brand safe and fraud free.” PubMatic, an SSP that appeared to categorically exclude selling MFA inventory in a June blog post titled “Advancing Responsible Media by Removing MFA Inventory From Auction Packages,” subsequently deleted that post and was reported by Adalytics to transact ads for 7,588 brands in January on MFA sites. “We continue to exclude sites that meet Jounce Media’s definition of MFA from PubMatic-sold auction packages and [private marketplaces] in the U.S. and always adhere to a buyer’s request for the inventory included in the deals,” the company said in a statement. “We removed the blog post that outlined this plan after receiving client and internal feedback about regional and channel-specific nuance that was not appropriately captured in the blog content.” Sharethrough, another SSP, also sounded like it was forswearing MFAs in June, days before the ANA issued the first part of its report. “Sharethrough becomes the first ad exchange to remove MFA sites … from all its off-the-shelf deals and custom PMPs created after July 1, 2023,” the company said in its June statement. Nevertheless, Adalytics reported Sharethrough was involved with MFA deals for 36 brands in January. “There are still several reasons why some MFA sites can occasionally appear on PMPs,” said Curt Larson, chief product officer of Sharethrough, in a statement. “One reason is that advertisers may disable the default MFA filter if they choose. Additionally, some advertisers are still using PMPs created before July 1, 2023. Another important consideration is that MFA sites are classified by our third-party partner. Since the MFA domain list is updated daily, there can be ad impressions delivered on websites before they are recognized as MFA and subsequently excluded from PMPs.” Index Exchange also has appeared to categorically reject MFAs, most recently in a presentation by Jessica Breslav, chief commercial officer of the SSP, at the ANA’s March 7 “Meet the SSPs” event. “We prohibit MFA content,” she said. “It’s something that we’re very proud of. And throughout 2024 we will continue to take a really proactive approach in auditing our exchange.” Nevertheless, Adalytics reported numerous MFA ad placements from Index Exchange in January. The reason is that sometimes sites that don’t qualify as MFA when they’re onboarded to the SSP change their practices later, and it can take some time for audits to catch that, said Jason Cicchetti, general counsel at Index Exchange, who heads up the compliance and quality department, in an interview. “MFA is prohibited on our exchange,” Cicchetti said. “No ambiguity there. But like with any policy, where the rubber meets the road is how do you enforce that policy.” Every one of the more than a million bundles and domains transacted on the exchange is reviewed every day, Cicchetti said, across a host of quality indicators. But one thing Index Exchange can’t control is when publishers initially “pass our sniff tests” but “later change in some capacity,” such as increasing how much traffic they pay for or how many ads they serve per page. One of the sites noted by Adalytics in January already had been removed from Index Exchange supply by February, before Index Exchange heard about the Adalytics report, Cicchetti said. “When we find MFA, we terminate it. That’s a fundamental difference between us and our peers.” Nuanced responses Nuance is a word heard among some SSPs explaining why MFA sites end up in their inventory despite their efforts. It may also apply to what many advertisers want. DoubleVerify last month captured the nuance with a three-tiered offering (high, medium and low) for pre-bid screening of sites based on how many characteristics of MFA they exhibit. IAS added a bit more nuance still last week with an offering meant to differentiate MFA sites from those that just have a lot of ad clutter. Bill Duggan, group executive VP at the ANA, said Adalytics confirmed that MFAs still exist, which he knew. But he said that the study wasn’t designed to make year-over-year comparisons or quantify the total percentage of MFA buying in the marketplace. “My expectation is that, among the advertisers in the ANA study, their MFA activity has declined,” Duggan said, adding that the ANA feels more advertisers should work from inclusion lists, not exclusion lists, to minimize MFA exposure. He added that the ANA is working with TAG TrustNet to create an ongoing programmatic benchmark on key metrics, including MFA, to provide a barometer on progress—or lack thereof—that it expects to release this summer. How the study worked Adalytics identified 21 websites whose traffic sourcing, content and ad delivery were consistent with the Association of National Advertisers, 4A’s, Incorporated Society of British Advertisers (ISBA) and World Federation of Advertisers joint definition of Made for Advertising websites. The sites sourced traffic via Taboola, Outbrain or Meta paid ads and got a high percentage of their audience from such paid media sources, according to Adalytics. The sites also were confirmed by DeepSee.io and/or Jounce Media to meet their definitions of MFA sites. Some of the sites varied on how many ads visitors were served based on where they came from. Visitors who came to a site directly might see no ads or far fewer ads than those who came from clicking on paid ads, even on the same article page. Pages served to visitors from paid traffic sources often had a “noindex” meta tag, which keeps crawlers from Google or Bing from indexing them or people from finding the pages through search. A direct visitor to one of the sites, say a marketer checking log files post-campaign, might never see pages heavily stacked with ads that people who come from paid traffic sources did. Data about how the sites served ads came through an audience panel of volunteers using an Adalytics browser plugin, manual inspection, crawlers, data extraction from internet archives such as URLScan.iro or HTTP Archive and other sources, according to the report. For each ad collected, Adalytics stored and analyzed metadata and source code to identify supply side platforms or ad exchanges involved in the transaction, as well as the demand-side platform and ad agency involved, where available. Since the study looked at only 21 sites over the course of a month, the data isn’t necessarily representative of an MFA universe that includes thousands of sites. The 21 sites focused on in the study were chosen to focus on those observed to serve as many as 1,000 ads to a visitor during a single page view session, said Adalytics’ Franaszek. See all of Ad Age’s 2024 Agency A-List winners here

Inside the ‘made for advertising’ crackdown—and why it isn’t working

Featured Stories