Easily spotted on the mobile web: holiday ad next to plane crash story; Muslim dating ad next to KKK story; beauty ad next to domestic violence story; car ad next to emissions scandal story…
[This post originally appeared on our sister site ClickZ.com, but we thought it was so useful we wanted to share it here as well]
Let’s admit it we’ve all had an occasional giggle when we spot an ad prominently displayed next to inappropriate content on the web. But for advertisers this is no laughing matter.
An ad that is not displayed in a contextually appropriate environment is not only a waste of marketing budget, but is a potential embarrassment (if shared on social media) or, at worst, damaging to the brand’s reputation.
Research by inMobi (via eMarketer), July 2016, highlighted in this column on mobile ad fraud reveals that 26% of advertisers state that concerns over brand safety is preventing the take up of programmatic purchasing (buying ads on the fly via an ad exchange) of mobile inventory.
While this is considerably less than the 48% concerned about fraud/viewability, this is surely a matter the industry needs to address.
The examples pictured in this column were easily found on the mobile web. Try it yourself by selecting potentially controversial stories and waiting to see what ad loads (often ads load slower than the content).
The sites we have featured are not implicitly “toxic”, such as pornography or gambling, but prominent news outlets where some stories will deal with unpleasant matters from time to time.
No one would want these news stories to go unreported. But that doesn’t mean that a holiday company wants its ad next to a plane crash story; a baby food brand ad next to a child sex abuse story; a Muslim dating service ad next to a KKK story; a beauty ad next to a domestic violence story or a car ad next to a story on the emissions scandal.
The ad business calls this brand safety or content adjacency. The issue is a real one.
But preventing brand safety issues has become increasingly complicated because:
Direct relationships between advertiser and publisher have increasingly been replaced by a web of intermediaries including ad exchanges and ad networks companies.
Many sites are implicitly safe, but publish eclectic content that could any subject – i.e. news sites or news stories/user generated content (UGC) on social media sites.
Keeping track of ads displayed on mobile sites and – particularly – mobile apps presents a unique set of challenges, compared with desktop web.
As Kurt Hawks, SVP of cross device and video at digital ad targeting specialist Conversant, explains:
Mobile ad tech is still maturing and while the mobile web operates similarly to desktop display, the in-app environment is a very different and more complex tech ecosystem. A rise in the intermediaries needed to execute in-app has resulted in a more fragmented digital supply chain that is more difficult to monitor.
Text, image and video-based analysis have gotten more adept at assessing brand fit within webpages, but a lack of mobile standardization, particularly in the in-app environment, can hinder the effectiveness of these tools. For example, the VPAID (Video Player Ad-Serving Interface Definition) standard is still not universally supported within in-app environments yet.
This is the forth in our series of columns on mobile ad quality: see also:
Mobile ad fraud
Combatting mobile ad fraud
Mobile ad viewability
Brand safety issues easily found on the mobile web
Let’s take a look at some examples of what appears to be untargeted ads appearing next to stories on three mainstream news sites: Inquisitr, Washington Post and The Mirror; and/or via Google’s AMP (accelerated mobile pages) search results.
Please note that brand safety is subjective, we can only guess that the brand, the viewer and the publisher would deem these contexts inappropriate.
It is hard to imagine many mainstream advertisers wishing to be associated with a story about a man being arrested for murder, a KKK march in favor of Donald Trump’s victory, or pictured above a picture of a KKK ceremony. There were plenty of examples spotted on various journals – but this was the most controversial: a Muslim dating site alongside this INQUISITR story.
Are stories of funerals or air crashes ever a good context for advertisers? But surely no travel brand would wish to be pushing overseas holidays against the backdrop of a funeral of the president of a football club wiped out in an air disaster as spotted alongside this Washington Post story.
Child sex abuse is also a topic that most advertisers would consider toxic, but a brand, particularly one of the magnitude of Heinz is will be questioning how an ad for baby meals ended up topping this Daily Mirror story about alleged child sex abuse at Chelsea Football Club.
Just to be certain, we checked the stories pictured in this column with Melody Gambino, director of marketing at brand safety expert, Grapeshot:
Every one of these examples is poignant. The key to brand safety is not only ensuring you don’t put your advertisement in front of bots or on spam sites, but also that the content surrounding your ad is not offensive. This can be in a broad sense (terrorism) or something specific to your brand (emissions testing for VW, for example).
But no one can say for certain what is considered brand safe or not except the advertiser, explains Melody Gambino
Brand safety means different things for every brand and every campaign, so clarity about what it means for you is imperative to ensure you are getting the most from your brand safety partners.
Defining brand safety
In the useful New Rules of Brand Safety, Grapeshot gives the following definition for brand safety:
The term “brand safety” is notionally understood. It represents an environment that is fundamentally not hostile, will not cause perceptions of uncomfortable association or, worse, spur unwelcome sharing or commenting.
The costs of an error in this arena can mount quickly, from the need to craft defensive counter-messaging to lost sales. Only after the fact does the degree of damage become clear.
Dirty dozen of toxic content categories:
There are 12 different content categories which, according to the report, tend to be considered toxic and are routinely excluded by agencies and brand safety tools. These are adult, arms, crime, death or injury, online privacy, hate speech, military conflict, obscenity, illegal drugs, spam or harmful site, terrorism and tobacco.
Based on these criteria, the three stories pictured above – the KKK organizer’s arrest, the Chapeco funeral and the Chelsea abuse allegations – all appear to fail the toxic test.
Presumably a rape story, even when associated a classic film such as Last Tango in Paris, is deemed toxic content. But should it be considered universally inappropriate for all brands?
The following ads have no contextual relevance to the film or the news story, but is this an association that charities or luxury brands should avoid? N.B. if you perform a mobile or online search on this story, you will find plenty of other examples, these are not isolated cases:
World Food Program ad on the Chicago Tribune.
Rado Swiss watches ad on the com.
Mercedes A-Class ad on Sky News.
Targeted or untargeted?
Advertisers attempt to improve the impact of ads by targeting the ads at visitors based on the publisher, the visitor, their location, and contextual relevance of the story using keywords, images and metatags.
There is little evidence targeting in the Last Tango rape story, but the following three examples suggest contextual targeting at its most unfortunate.
The Volkswagen emissions story isn’t a universally toxic topic – but it is one that auto manufacturers, such as VW, should and do seek to avoid. The ad targeting software appears to have picked up on the keyword “Volkswagen” on this Chicago Tribune story to serve up ads for a VW leasing service. Unfortunately the ad targeting failed to pick out the word “emissions”, so did not put the break on the ad.
In the second example, the ad targeting seems to pick out the keywords “makeup tutorial” in this Mediaite story to serve up an ad for a Beauty app and a recruitment ad for models. To a computer this will look like a good opportunity, but no beauty brand will want to be associated with the use of makeup to hide domestic violence.
It is unlikely that the CBC house ad “Holiday like you mean it” is contextually targeted at this story 3 about drug use by Hitler and Nazi Army, but a reader would be forgiven for thinking so.
How brand safety works
There are two key methods for avoiding brand safety issues:
Blocking websites/URLs that are known to contain contentious material.
Detecting context from keywords either a) pre-bid, preventing the ad space being purchased at the exchange) or, where not possible, b) post-bid, where the ad is blocked from appearing on the page.
The first approach relies on agencies maintaining vast blacklists (and whitelists) of websites. Ad technology then blocks the ads from displaying on these blacklisted sites. The problem with this method is that it assumes that all content on blacklisted sites is bad for business and all content on whitelisted sites is good for business – which as we have seen above is not the case.
The second approach is more complicated. Ad quality/verification software, from vendors such as Integral Ad Science and Meetrics scan websites for possible issues for advertisers.
Jason Cooper, general manager, mobile at Integral Ad Science, tells how it works:
When web pages load they are requesting content from all across the Internet; text, copy, images, advertising content, tracking and analytics pixels among others. In each one of these requests, a little tout runs ahead declaring the page that is making the request; by intercepting that call, a verification vendor can decide whether that page meets the brand safety threshold of an advertiser before allowing the ad to serve.
However, apps don’t send URL’s, in fact identifying the app name requires a number of different strategies and is something that will be solved incrementally over time.
Issues with mobile (apps)
Felix Badura, director product development at Meetrics, explains:
Tips and common mistakes with brand safety
Don’t assume that because you are buying ad space on a known app it will be safe for your brand. Apps don’t need to be porn free to tick the box of ‘safe’. An article about performance enhancing drug scandals next to Reebok or Nike ads can be equally as damaging. – Melody Gambino, Grapeshot.
Brands must be especially vigilant and engage vendors that have robust brand safety capabilities (semantic analysis of text; tags; imagery and tonality of web pages and apps; video-level data), but must also ensure these partners are capable of handling the nuances of brand safety in-app. – Kurt Hawks,
Only serve ads via the big marketers like AdMob or MoPub and keeping the pressure high on insisting to have a third party measuring the data. – Felix Badura,
While it is a good idea, to retrieve and analyse the Package names of the apps where an ad is running, one should be aware that even problematic apps might be uploaded using inconspicuous names to stay below the radar. – Felix Badura, Meetrics.
This is Part 39 of the ClickZ ‘DNA of mobile-friendly web’ series.
Here are the recent ones:
Mobile ad viewability: what is it and does it matter?
How agencies and advertisers can spot and combat mobile ad fraud
Mobile advertising accounts for nearly half of digital spend, but it comes at a price: ad fraud
Where is Google heading with mobile local search?
Is Google killing mobile organic search
Andy Favell is ClickZ columnist on mobile. He is a London-based freelance mobile/digital consultant, journalist and web editor.
Contact him via LinkedIn or Twitter @Andy_Favell.