Hypebusters: Ad Verification is No Cure-All

To many brand marketers, the online advertising space is still a bit like the Wild West. New frontiers like ad networks, trading desks, demand-side platforms and real-time bidding only add to the confusion. As some marketers see it, they’re buying inventory on unknown sites, and they’re putting their faith in the network, exchange or agency to deliver their ads to consumers in brand safe environments. Enter the ad verifiers.

This is a category that seemingly sprung up overnight. It even got its own box on the Luma Partners slide. Unfortunately, pure-play ad verification, as it stands today, is still a reactionary tool, rather than a proactive — and truly protective. Brand marketers using ad verification might still run into a host of safety issues.

One of the major challenges with the current iteration of ad verification tools is that they determine safety on a very binary scale, where sites are either safe or not. As John Ebbert of AdExchanger pointed out earlier this year, verification still produces lots of false positives. His comical example is “This wasn’t adult content! It was Obama at the beach!” but the process is a lot more complicated. The definition of safety is not binary, and each advertiser has its own idea of what safety means based on their brand message.

Take women’s interest sites. Most have sex advice channels on their websites — take a look at some of the racy headlines on Redbook’s Sex and Love channel. Despite that language, brands like Levi’s and Zipcar serve ads on the page. Some brands may object to this, but others clearly do not, and binary ad verification doesn’t always see the difference.

Many verification companies introduced sliding scales, but these tend to over-complicate the process. It’s very difficult to decide if something is only partially pornographic or profane. Can you really assign a numerical grade to those topics and expect to make advertising safe while still achieving scale? Safety should be simple and effective, if nothing else.

I mentioned before that ad verification is often a reactionary safety feature, rather than a proactive one. Ad blocking software can prevent ads from appearing on inappropriate pages, but it’s still not 100 percent effective, especially with new ad serving strategies like real-time bidding. RTB is one area where safety is needed the most, because advertisers buy impressions unaware of the URL. If an advertiser buys an impression next to questionable content, the software blocks the ad. The problem is, the advertiser already paid for the impression.

The flip side of that is that verification without ad blocking only tells the advertiser what happened after the fact. So, when applied to RTB, ad verification tools can leave an advertiser paying for an impression where no ad is served, or paying for an ad that already ran in bad location. Additionally, verifying ads blocked by the same company is an inherent conflict of interest. That’s a lose-lose situation.

Above all else, most safety solutions are not truly-page level, which means they can’t process the detailed information you need for true ad verification. They are domain level classifications applied at the page, making for a blunt instrument in determining safety. Sometimes the same word has two meanings. It’s difficult for people who speak English, and equally confusing for non-semantic software. We just conducted a study based on the keyword Joplin. A few weeks ago, most of the Joplin content running through our system pertained to the late singer Janis Joplin, which is a good environment for entertainment advertisers.

A few weeks later, a tornado hit the town of Joplin, Missouri. Suddenly, most of the content containing the word “Joplin” centered on the tragedy, damage and the aftermath. Content with the word “Joplin” is no longer appropriate for a record company advertising Guns N’ Roses’ “Appetite for Destruction.” I don’t mean to be flippant about this tragedy, but it’s a clear example of how non-semantic verification software doesn’t know the difference.

Brand safety is a crucial part of online advertising, and that’s not up for debate. If brand advertisers see the Web as the Wild West, there’s no chance they’ll increase their share of digital spend, especially in RTB. Ad verification has its place, but there’s still a chance that brands will encounter safety issues even after employing verification and blocking software. For brands to really start shifting budget online, we need to guarantee safe environments through semantic analysis. Black and white, binary URL systems just don’t cut it anymore.

Andy Ellenthal is CEO at Peer39, a semantic ad targeting firm.

https://staging.digiday.com/?p=3117
Digiday Top Stories
Ad rendering preventing in staging

Ad position: web_bfu