New Tool First to Leverage Image and Audio Recognition; Allows Ad and Media to Perform 2.5X Better than Standard Topic Targeting for Companies Developing Brand Safety Measures
Fullscreen, the global entertainment company serving social-first experiences to creators, brands, and consumers, announced the launch of Fullscreen Shield, a proprietary, in-house brand safety product now available widely. Fullscreen Shield is the first social cross-platform solution designed to provide brands with another layer of safety against negative online video content.
Fullscreen Shield is the first customizable product for videos that is both thorough and transparent, leveraging both image and audio recognition in addition to analyzing signals from metadata, tags, titles, comments, and audience ratings. This data collected is then audited against Fullscreen’s corpus of known unsafe terms, while its machine learning capabilities — which are informed by human insight — discover new unsafe words in real-time as the landscape evolves. An easy-to-understand score is generated based on the weight of signal triggers, giving brands results that are bucketed into ratings groups.
“Brands today face a number of challenges in maintaining brand safety for online video content. Five billion videos are watched each day, and the industry lacks the tools needed for a simple but comprehensive way to automatically screen content,” said Mark Williams, senior director media operations, at Fullscreen. “Fullscreen Shield is an industry leader in its ability to scan content, as it’s the only tool that can review every dynamic visual identification and all audio, rather than just a thumbnail image.”
Fullscreen Shield flags videos that are potentially unsafe for brand association by assigning video ratings that trigger human reviews prior to launching media campaigns, ensuring that advertisements appear only in safe environments. Fullscreen Shield takes a unique approach to brand safety, identifying campaign-specific content that provides relevancy and recency for optimal brand performance. This type of contextual and audience-based targeting allows advertisements to perform 2.5 times better than when using standard topic or interest based targeting. Shield currently contains more than 25 categories that map against more than 12,000 keywords, a number that is constantly expanding.
“Brand safety is not universal – there is no single list that incorporates all unsafe content as each list is brand-specific, which is why Fullscreen Shield combines machine learning and human review, resulting in the industry’s most comprehensive brand safety tool,” Williams continued. “We want to ensure that our product can be customized to each brand’s specific needs, and that it is constantly evolving to be relevant for all types of content, whether that be animated or video game content or Facebook and Instagram videos.”