Contextual targeting is undergoing a significant transformation with the integration of generative AI. Gone are the days when media planners relied on basic audience segments to target consumers. Today, a recently launched platforms for editorial ads allows marketers to enhance their strategies by simply inputting a creative brief into an AI-driven interface, which then generates a list of relevant URLs for ad placements.
Startups like Cognitiv and RTB House are at the forefront of this evolution, utilizing generative pre-trained transformers (GPTs) to refine contextual segments. These innovative tools harness natural language processing to improve the relevance of ad placements, addressing the common pitfalls of traditional contextual targeting. For instance, Aaron Andalman, Cognitiv’s co-founder and chief science officer, points out that the existing contextual offerings can be overly broad. Marketers often find themselves ticking boxes to target generic topics, such as “articles about baseball,” without considering the nuances of content relevance.
Conversely, keyword-based approaches can lead to overly restrictive targeting, as Jeremy Fain, Cognitiv’s co-founder and CEO, explains. Keyword blockers might prevent all content containing specific words, such as “cheat,” without distinguishing context. Cognitiv’s sentiment model addresses this by allowing brands, such as Nike, to filter out negative uses of “cheat” while still allowing positive mentions.
RTB House has also embraced generative AI in its approach to audience segmentation. Through its PrimeAudience platform, launched last June, the company develops contextual segments by analyzing content that aligns with marketers’ goals. For example, if a marketer wishes to target individuals planning a trip to Paris next summer, PrimeAudience’s algorithm identifies relevant articles discussing seasonal events and local attractions.
How It Works: AI-Driven Contextual Targeting
Both Cognitiv and PrimeAudience operate on a prompt-based system. For instance, marketers looking to reach consumers interested in Yeti cups can paste relevant content or briefs into the platforms. This allows them to modify their prompts and refine results until they meet specific criteria. Cognitiv’s tool provides real-time feedback, enabling marketers to accept or reject individual URLs, thus streamlining the process of ad placement.
Moreover, Cognitiv’s AI allows for custom filtering based on sentiment and inclusivity. Marketers can define their desired sentiment—positive, neutral, or negative—while the tool evaluates articles for their alignment with inclusive language standards.
PrimeAudience offers a similar experience, guiding marketers through the process of audience creation. Users describe their target audience, and the platform classifies articles as “Yes” or “Probably” based on the content’s relevance. The algorithm also provides insights into the likelihood of article readers being part of the desired audience, offering marketers a comprehensive view of potential placements.
For both platforms, the end goal is to generate a list of matching URLs for targeted advertising. These URLs can be integrated into Demand-Side Platforms (DSPs) for ad buying. Cognitiv partners with major supply-side platforms like Magnite and PubMatic, while PrimeAudience collaborates with partners such as OpenX and Microsoft.
Privacy and Data Considerations
As contextual targeting moves toward more sophisticated solutions, privacy remains a key concern. Both Cognitiv and PrimeAudience have implemented measures to protect user data. PrimeAudience employs pseudonymous IDs, while Cognitiv emphasizes its cookieless, ID-less technology to eliminate privacy concerns altogether.
Before deploying these generative AI tools, companies must first establish extensive content libraries. Cognitiv scrapes millions of web pages daily to feed its AI models, while PrimeAudience sources its data from bid requests, ensuring compliance with user consent standards. Both companies leverage a combination of open-source large language models (LLMs) and proprietary technologies to power their contextual targeting solutions.
The Future of Contextual Targeting
As the industry prepares for the decline of third-party cookies, the interest in AI-driven contextual targeting tools is expected to rise. Cognitiv reports significant engagement from consumer packaged goods brands, while PrimeAudience anticipates adoption from various sectors as marketers increasingly recognize the advantages of precise, AI-enhanced targeting.
With generative AI making contextual targeting more accessible and effective, marketers can expect a smoother journey toward reaching their target audiences with relevant and timely content.
