| ALAN WOLK (AW): Why is contextual targeting taking off now—what are some of the big issues it is solving for? RAGHU KODIGE (RK): Recent advances in AI technology have finally made precise video content analysis possible at scale. While contextual targeting has proven successful in digital advertising, it's only now technically feasible for video content at the scale required for CTV. The industry is moving beyond basic capabilities like genre-based targeting or text-based programming information. At one point, just showing different ads based on location was innovative for CTV. Today's advertisers demand more sophisticated approaches that maintain privacy while delivering relevant experiences. This solves two critical challenges: creating an integrated viewer experience by aligning ads with content moments, and giving advertisers more sophisticated targeting capabilities. As advertisers gain better insight into campaign effectiveness through improved data availability, they increasingly recognize that contextual relevance drives superior results. AW: What is "multimodal AI" and how does it help to set Anoki's approach apart? RK: Multimodal AI simultaneously analyzes multiple types of information within video content - dialogue, music, visuals, and scene progression. Our technology stands out through scene-level understanding, processing these signals concurrently to understand context just as a human would, but at scale. We extract and interpret every meaningful element - from visuals and audio to metadata such as objects, locations, emotions and activities. This enables powerful capabilities like cross-modal match, where advertisers can find relevant scenes using text queries, video assets, or audio clips. For example, a pet food brand can use scene-level targeting to match ads with positive programming featuring a beloved family dog. This granular analysis delivers more precise and relevant ad placements than solutions that only analyze content at broader levels. AW: What sort of taxonomy do you use to classify emotions and scene-level details? RK: Our approach transcends fixed taxonomies, using AI that understands contexts through rich semantic representations. While we maintain standard emotional classifications, we're not limited by predefined categories. The system can identify specific contextual elements like "tearjerker moments" or "scenes of women applying makeup" based on advertiser needs. We also incorporate brand safety classifications while maintaining flexibility to adapt to any specific targeting requirements. This combination creates more nuanced and accurate scene-level opportunities. Our semantic approach allows us to understand complex emotional contexts and scene characteristics that traditional rigid taxonomies might miss, creating more sophisticated targeting possibilities. AW: How is Anoki addressing the issue of overfrequency? RK: By targeting specific contextual moments, we create more varied ad exposure patterns. Instead of repeatedly serving ads based solely on audience identification, we deliver them when viewers engage with relevant content in the right context. Through our AdMagic technology, we dynamically optimize creative elements to maintain engagement even in cases of multiple exposures. This doesn't replace traditional audience targeting but rather enhances it by adding contextual relevance, creating a more balanced approach that optimizes frequency while maintaining effectiveness. The combination of contextual targeting and creative optimization helps advertisers maintain their desired reach while significantly reducing audience fatigue. AW: What metrics are most important to your clients? RK: Our clients' key metrics align with their broader marketing objectives - brand awareness for brand advertisers, conversion metrics for performance-oriented campaigns. Through A/B testing, we measure contextual targeting's incremental impact against traditional approaches, allowing advertisers to quantify the value of reaching audiences in relevant moments. This works within existing measurement frameworks while improving outcomes across standard metrics. Whether focused on brand lift, engagement rates, or conversion metrics, our technology helps optimize for these goals by ensuring ads appear in the most relevant and impactful contexts. AW: Do you see contextual targeting becoming the dominant form of ad placement on CTV? RK: Just as no advertiser would place advertisements for baby products in a bar, contextual advertising fundamentally matters in CTV advertising: it makes audience targeting work harder. It works alongside other strategies and makes them more effective – that’s the beauty of it. For viewers and advertisers alike, it will always be better to match advertising with the right context. So this isn't about choosing one approach over another; it's about recognizing that finding the right contextual environment for an ad maximizes its impact. Through advances in multimodal AI, the technology now exists to ensure ads appear with scene-level contextual relevance at scale, similar to standard practice in digital advertising. As the technology continues to mature, we expect the synergistic relationship between contextual and audience targeting to become an increasingly crucial component of effective CTV campaigns. AW: How can your approach to contextual work with shoppable and other interactive formats? RK: Contextual targeting amplifies the effectiveness of interactive formats by ensuring they appear in relevant moments. Viewers are more likely to engage with a shoppable ad when it appears in a contextually relevant scene - like a cooking-related overlay during a culinary scene. Rather than simply adding interactive elements to any placement, we ensure these features appear when viewers are most receptive, creating a more natural shopping experience that drives higher engagement. This natural alignment between content and commerce creates more organic opportunities for interaction, significantly enhancing the performance of interactive ad formats. AW: Do you have any plans for international expansion, and how will you handle cultural differences? RK: We're expanding internationally, starting with Australia and Western Europe. Our strategy leverages multilingual AI models trained on local content to naturally capture cultural nuances and context. The technology allows for market-specific customization, fine-tuning models for regional sensitivities while giving advertisers granular control over brand safety parameters across markets. This enables both global standardization and local customization, ensuring we effectively serve diverse markets. Our flexible approach allows brands to maintain consistent global standards while adapting to local cultural norms and regulations as needed. |
