“On TV And Video” is a column exploring opportunities and challenges in advanced TV and video.
Today’s column is written by Sam Appelbaum, general manager at Yellowhammer Media Group.
EMarketer predicts that nearly three quarters of all display advertising will be transacted programmatically this year. Programmatic video isn’t far behind, but the premium in price it commands, coupled with complexities in how the inventory is classified, makes video especially susceptible to fraud.
An inordinate amount of time and money is focused on leveraging technology to solve for marketers’ brand safety concerns for video and display. Automated technologies to detect bots or viewability are unable to catch a particular type of video ad fraud: misrepresented video-ad classifications.
Video inventory is available across a variety of different types, from click-to-play and audible in-stream placements to autoplay muted content. In many cases, humans are responsible for labeling video type classifications that ultimately bubble up in the bid requests to downstream demand-side platform buyers. These classifications have a direct impact on the price a marketer is willing to pay for a particular impression.
Too Much Room For Error
In display, the universal ability to execute third-party JavaScript allows for sellers to be less reliant on human inputs for inventory classification. This means there is less that a buyer has to trust since technology can largely detect the relevant targeting criteria, ad size, device or geography.
Video, however, with its reliance on VPAID to navigate the complexities of disparate player technologies and content types, requires more manual inputs. An SSP will require its publishers to designate settings, such as stream position, audibility and user initiation, for each video placement within the exchange.
These settings affect the desirability of that placement and, with it, the ultimate price that it commands on the exchange. Buyers make decisions based on the price difference between these details. For instance, Facebook elected to force audio-on for all videos in a user’s news feed because of what premium brands are willing to pay.
Sites with less scrutiny than Facebook may be tempted to mislabel these classifications to increase their profits. SSPs and networks don’t have much incentive to audit how video inventory is classified because they profit, too. Many actors understand that at the end of the day, it is difficult for a brand to truly verify what exactly they are buying.
As with any manual process, mistakes are unavoidable, but in some cases, these misclassifications are intentional. While SSPs do routinely reprimand or kick out publishers if they go too far with misrepresenting video inventory, incentives to be proactive are low unless buyers make it clear that they’re paying attention.
Combating Cat-And-Mouse Fraud Technology
So, what steps can marketers take to mitigate their exposure to video inventory types they ultimately do not want to buy?
Step one is to create downward pressure on the community writ large to further automate the verification of these key classifications. The introduction of JavaScript in the VAST 4.0 specs is a step in the right direction. The brand safety vendors also seem to be uniquely equipped to help solve this problem.
The technology will take time to develop, which might pose problems for an industry as fast moving and dynamic as programmatic. For example, verification in non-VPAID compatible inventory types like in-app mobile and connected TV have proven to be a thorn in the side of verification companies that rely on VPAID to execute their services.
As programmatic TV and over-the-top devices continue to proliferate, these issues will only become more difficult to solve.
Find The Quality Within The Quantity
Until every aspect of programmatic video advertising is fully automated, these issues will persist. However, rather than admitting defeat and becoming suspicious of every partner, brands need to take a bigger role in fraud prevention and be proactive about correcting genuine mistakes. To do both effectively, they need to match humans with humans.
Brands should ask partners for the account IDs of the media companies they buy with, including SSPs, publishers and networks. Brands should try to contact the people behind the account IDs and ask how they categorize inventory and what steps they take to diminish errors and assure brand safety against fraud.
Of course, brands will never be able to meet with 2,000 websites directly, but they can work actively to spot-check publishers and vendors, build strong relationships with their most important partners and discuss their intentions to dramatically decrease issues with video.
Even with automated brand safety solutions, marketers would be well served relying on fewer trusted partners that are transparent about their operations. This type of personalized media buying offers quality that can be trusted even as the category grows and changes.
Heineken, for example, has successfully focused on fewer, more transparent direct relationships in video, and it is not afraid to stop working with partners that don’t take video fraud seriously.
In this way, brands can pick up where verification technology leaves off by engaging with a trusted circle of video providers that they can get to know and meet with regularly. This informed and transparent accountability by the marketer will produce positive lasting effects on the success of their programmatic campaigns.
Follow Yellowhammer Media Group (@yellowhammermg) and AdExchanger (@adexchanger) on Twitter.
