Brand safety: Avoiding fake & hyperpartisan news on the Google Display Network
Thanks to programmatic targeting, brands can find their ads displaying next to editorial and other ad content that may not be in keeping with their values and messaging.
What does online brand safety mean in the age of hyperpartisan, sensationalized and fake news content?
With programmatic targeting and retargeting, brands can find themselves cozied up to content that may not match up with their values and messaging. That’s not new, but the proliferation of sites aimed at attracting clicks by appealing to humans’ basest need to have their viewpoints validated adds another layer of complexity for brands buying ads on open ad exchanges and ad networks like the Google Display Network.
Well over 1,000 advertisers have reportedly pulled out of Breitbart, and yet that site is part of the Google Display Network. So are hundreds of other hyperpartisan political sites — both left-wing and right-wing, with content that ranges from flirting with untruths to outright lies — on which brands might be surprised to find their ads appearing (more on that below). Here we are looking at websites only, but ads can also appear alongside YouTube and apps should also be considered.
“Fake news” definitions can vary widely, from misinformation to conspiracy theories, to hoaxes to blatantly false reports. Google’s take as it pertains to the GDN is different still. Google has no publisher policy against sites running false news stories, as long as they aren’t misrepresenting who they are or the intention of the content — i.e., sites can’t impersonate a news outlet, and news headlines can’t link to diet pill promotions. We’ve written about what exactly the policy covers and doesn’t in our companion piece, “Google isn’t actually tackling ‘fake news’ content on its ad network.”
This can be awkward. Here are some examples:
Smartfood “You deserve delight” ads appearing next to NSFW headlines.
Planned Parenthood “Give Now” ads showing up on a far right-wing website that hasn’t even bothered to fully update its “Sample Page” template.
Bergdorf Goodman and Alliance for Healthcare Security ads on shock-pundit AnnCoulter’s site.
How does this happen?
There are several ways ads can show up on sites in the Display Network. First, these placements could be intentional — with the sites chosen specifically as “managed” placements on the Display Network by the advertisers. That’s not likely the case here, but it’s possible.
The other scenarios rely on automation. Google has added more targeting controls to AdWords, and even more in DoubleClick, for the Display Network from the original contextual targeting with keywords (e.g., ads for health insurance plans can appear alongside articles about the ACA). Other targeting options include demographics, interest and affinity audiences, topics, and of course, retargeting to customers and/or past site visitors.
The ads in the examples above were not served as the result of retargeting campaigns. The Bergdorf Goodman ad shown above could be the result of interest targeting, which can include in-market (like “Women’s Apparel”) or affinity audience targeting (like “Fashionistas”) based on my previous browsing behaviors, including visits to other high-end retail sites. Google’s interest targeting is built on a mix of third-party data and browsing behavior on pages, apps, channels, videos and content on YouTube and the Google Display Network sites.
Topic targeting isn’t as simple as targeting websites that cover specific topics such as Right-Wing and Left-Wing Politics, which are both options. Topic targeting occurs at the time an impression becomes available and is based on the content on the page, not the overall political vantage, for example, of the site overall. That means advertisers targeting Right-Wing Politics can find their ads shown on sites that skew left-wing if the content includes keywords that signal a right-wing context, and vice versa. For that reason, the topics that are listed with websites in the Display Planner will not necessarily correlate to the types of content ads can display against with topic targeting.
Retargeting may be the trickiest decision for brands to weigh when considering where their ads should appear. Here are several examples of retargeted ads for retail brands appearing on far-left- and far-right-wing sites. These ads happen to be for retail brands, but Marketing Land also saw examples from CPG, telecommunications, automotive and financial services advertisers, among other verticals, while reporting this story.
The ad company you keep
Often, it may not be the content but the content recommendation ads and ads from other networks on a lot of the sites that are more unsettling to brands. Many of these sites are absolutely safe to visit, others less so.
Some mobile experiences include pop-up ads and content recommendation ads that display between an article title and body.
One of the worst examples of ad loading I saw was on a site called LeftLiberal.com. The page below had 75 ad trackers on it. There is a Google-served banner at the top, a block of sponsored links from Content.ad, a virus alert ad in the bottom right, and a video ad served by Epom in the bottom left that is covering up another Google-served responsive text ad. This site was also one of the many using popups Dingit.tv.
I should also offer a consumer warning here: some sites I visited while reporting this story automatically opened a new browser tab with phony virus protection alerts designed to be hard to close out of or had sketchy virus warning ads appear in the bottom right corner of the page. The worst of these scenarios opened up a new browser tab automatically with a “safety alert” that claimed my Facebook login, credit card number, email account login and photos had been compromised and that I needed to call support. (Do not
call or click on these kinds of ads.) Several sites also triggered background pop-up video ads.
A site called Prntly.com that dubs itself “America’s Top News Site” took nearly 30 seconds to load and had 66 ad trackers on the home page alone.
Marketing Land found and reported to Google several examples of apparent violations of the Misrepresentative content policy, and we have asked questions about violations of the Valuable content policy that covers ad-to-editorial content ratios. We have also asked about whether certain examples violate Google’s AdSense hate speech policy that prohibits content that advocates against an individual, group or organization. We will update here when we hear back.
Does environment matter?
Somehow, in the age or programmatic ad buying, context seems to have taken a back seat to audience reach for many brands. When aiming to reach certain audiences or retargeting, does it matter if a brand’s ads show on content it wouldn’t otherwise target if that’s where their site visitors and customers are going?
Different brands will have different answers to questions about context and environment. When brands set impression and audience quotas, they and their media buyers may be reluctant to cut of significant sources of reach.
But in this new landscape, marketers running campaigns on ad networks like the GDN need to be asking these questions for themselves, because Google won’t do it for them.
How to opt out
Many companies will have no problem with their ads appearing next to hyperpartisan content or on sites with fake news. And some may want to explicitly target audiences reading this type of content.
For AdWords users that want to opt out, there is no Site Category exclusion option for political content, but there are other options — none quite perfect.
Placement exclusions are the primary way Google suggests addressing opt-outs. Advertisers can exclude individual sites and even individual pages on which they don’t want their ads to appear. This can be hard to manage when there are new sites coming into the GDN on a regular basis, and in many cases, advertisers won’t know to exclude a site until after their ads run and the sites show up in site placement reports. Another thing to note in terms of brand safety is that AdSense publishers have the option of making themselves anonymous to advertisers. These sites show up as anonymous.google in advertisers’ placement reports. It is possible to exclude all anonymous.google sites, though this can be a blunt instrument that excludes brand-appropriate sites and even converting sites.
Topics exclusions are also a blunt tool. There are several granular options under Topics, including Right-Wing and Left-Wing Politics, but as discussed above, this targets content not sites. Excluding Left-Wing Politics could mean ads won’t show next to right-wing content.
Topics exclusions can also cut off access to content that isn’t necessarily extreme, too. Similarly, interest exclusions can paint too broad a brush in many cases.
Advertisers can also add contextual exclusions by adding negative keywords to their display campaigns.
I’ve compiled a Google Sheet of some sites included in the right-wing and/or left-wing politics topics in Display Planner (reminder these topics don’t align to Topics targeting in campaigns and the list is always subject to change). There is a mix of mainstream, left-wing and right-wing sites. It’s meant to give advertisers a sense of the political content inventory in which their ads might appear on when running GDN campaigns.