Nuance. That’s what’s been missing in the quest for brand safety over the past several years, says Mark Pearlstein, chief revenue officer at 4D, the Context Outcomes Engine at Silverbullet, which defines itself as “a product and services business for the new marketing age.”
“In the era of cookie targeting,” Pearlstein says, “the mantra of brand safety was to ‘avoid any content that could potentially harm my brand.’” While Pearlstein acknowledges that this approach was universally acclaimed as logical when ads were served based primarily on cookie profiles, he believes that, in a soon-to-be cookie-free world, the traditional approach to brand safety is now much too over-arching. “What we’ve come to realize,” he says, “is that if you draw the lines of what’s safe too broadly, you end up rejecting and not monetizing a lot of good content and, in the process, missing key moments that may actually be perfect for brands.”
Take content about alcohol, for example. In the binary world of brand safety, Pearlstein says, any content related to alcohol would be deemed negative and therefore blocked by brand safety tools. That action would make sense, he points out, if the content covers alcoholism or liver damage or drunk driving. “But how about sipping cocktails by the beach at sunset?” he asks. “Not every brand wants to avoid that. There clearly are shades of gray in terms of what’s acceptable and what’s not.” And those shades of gray, he suggests, are what draw the line between brand safety and a more balanced and multifaceted appraisal of brand suitability.
Pearlstein cites the work of the Global Alliance for Responsible Media (GARM), an arm of the World Federation of Advertisers. As he describes it, GARM “started out saying, ‘We need to subdivide a lot of these categories that we often called inappropriate and that we’ve been blocking and finer tune them, putting in variations and gradations.”
Learning to See Shades of Gray
Though to many in the industry, Pearlstein says, GARM’s work looked like a way to define brand safety better, he sees it as an indication that the industry is moving away from rigid brand safety frameworks toward more nuanced brand suitability.
Fueling this move over the past 18 months, he says, have been not only the upcoming demise of third-party cookies, but also a realization by the industry that “balance needs to be reestablished between publishers, media sellers, and media buyers. Suitability—or, more accurately, suitability targeting—actually takes the inverse approach to ‘brand safety blocking.’ While that approach is about identifying anything that could be bad and removing it, suitability targeting is about finding all those places that are good and pinpointing those that are bad.” That means, he adds, “it’s no longer about the high-level content classification of a topic, but instead it’s about how the topic is being framed in the context of the brand, what it is they’re advertising, and the moment they’re trying to identify.” So if the topic is alcohol but the context is liver damage, the suitability of the environment is measured differently than if the topic is alcohol but the context is the best summer drinks to have on the beach.
What this involves, Pearlstein explains, is the development of a very different framework than was possible with legacy brand safety tools. Those tools were never intended to pull positive context, negative context, sentiment, and other data points into a single definition. But looking carefully for suitability—for the right moment for the right ad—involves, he says, “analyzing each page holistically against advertiser-specific criteria,” a task that requires purpose-built technology. It also, he notes, is based on information drawn from a network of information providers that can assemble and layer these various data points, helping a brand understand the perfect time and place for an ad to run.
What COVID Taught Us
The importance of looking for gray areas, for probing the range of suitability measures, he notes, has been particularly evident this past year, given the spate of COVID-19 coverage. “Fundamentally,” he contends, “as a brand marketer, you could have been put out of the revenue business if you said that everything related to COVID needed to be blocked.”
The suitability targeting approach, however, looks at something like COVID coverage through a much finer lens. “On the one side,” Pearlstein says, “COVID articles that talk about death are places where most brands don’t feel comfortable advertising.” However, he adds, “look at all the COVID articles focused on the vaccines, the optimism, the hope, and all the feel-goods that come from that. That’s a great example of where suitability targeting versus brand safety avoidance can make the difference.”
By substituting suitability for safety, Pearlstein says that the goal is “to take a glass-half-full approach to the world based on the belief that content is fundamentally good and that you can just remove those nuances that are bad.” To do this, however, you need a blended solution that looks at everything at the same time. “In other words,” Pearlstein explains, you need a tool that’s “built to identify those areas that are relative to your brand and that can then fine-tune, pinpoint, and remove those instances that are off-brand.” By definition, this means that a model that allows for suitability targeting would require what Pearlstein sees as a “much different set of technical requirements that need to be built and brought to market for the benefit of advertisers.”
Moving Beyond Keywords
One such model was explored recently in a study conducted by Factmata, an artificial intelligence company focused on the internet, along with SilverBullet’s 4D division, which offers brands access to a wide range of data specialists through its Dimension Marketplace. In this study, Factmata looked to move, as its CEO, Dhruv Ghulati put it, “beyond keywords and labeling of standard brand safety violations.” By pairing AI with feedback from a variety of relevant communities such as journalists and advocacy groups, the company scored content against eight different “signals”—racism, hyper-partisanship, fake news, sexism, personal insult, threatening language, toxicity, and obscenity. The study’s results reinforced the need for a blended approach to determining brand suitability, couched against the conviction that not only is it important to move beyond the scope of traditional brand safety methods to determine suitability but that, as the study put it, “modern marketers who want to thrive in the new marketing age need to be aware of nuance and true context … in order to align with the right message.”
And moving toward this “blended” approach has financial implications as well. Factmata has found that unsuitable content flagged through their study but missed by existing brand safety vendors was equivalent to 0.71% of total spend. Extrapolating that to current global spends on programmatic advertising, Factmata estimated that advertisers in 2020 may have spent $898 million against unsuitable content.
This, of course, plays both ways: As the Factmata study shows, it’s possible that suitable-seeming content can, when analyzed more thoroughly, actually be unsuitable for brands. On the other hand, as Pearlstein’s liquor and COVID examples prove, there are many times when content that seems unsuitable when chosen strictly by keywords is actually a perfect placement for an advertiser. And both examples reinforce the importance of using suitability targeting to cast a net that is wide enough to identify the right moments at the same time that it protects against an increasing array of threats.
Putting Reach Within Reach
The limitation of current tools with which not only safety but suitability can be assured has already begun to resonate with marketers. A 2019 study by CHEQ and Digiday showed that two-thirds of respondents were skeptical about the brand safety tools with which they were familiar. At the root of this concern was reach: 92% said that not achieving adequate reach would push them to forego the use of these tools entirely; almost all respondents said they were seeking customized tools, with an eye toward both safety and reach.
The point, Pearlstein concludes, is that the quest to balance brand suitability with reach is complicated and steeped in nuance. “The number of signals today that we can glean from the digital experience is tremendous,” he says, as is “the number of companies that are identifying those signals to build datasets.” Moreover, he adds, while it’s important to have access to those data specialists that are providing those data points, it’s equally true that “not every data point may be relative for every marketer in defining suitability,” As a result, he says, “it’s critical to have a marketing model that says, ‘I can quickly and easily plug in specialists in new areas that are innovating to identify new data points that are relative to my brand.’” This, he points out, was Silverbullet’s goal in developing 4D. The data specialists that work with 4D, “allow brands to capture such data signals as page sentiment, weather, and domain quality to pursue in-the-moment marketing,” and 4D’s technology allows brands to determine where and when that data is relevant to their campaigns. This type of approach, Pearlstein says, underscores the nuances inherent in the effort to define brand suitability. Each brand, each ad, each situation, each context is different. Only by having access to the right data, and knowing how to analyze it, can real suitability be determined.
In the next installment of this series, we’ll redefine “contextual” as a tool in the search for brand suitability and will investigate the ways in which brands, with help no longer available from third-party cookies, can leverage first-party data—and evolve, using the new methodologies that first-party data is enabling. `