We understand that effectively reaching target audiences remains a key concern for marketing leaders. In fact, over half (53%) of executives worldwide say their leading concern regarding their digital advertising strategy is being able to reach target audiences effectively at scale (source).
Advancements in AI, particularly Large Language Models (LLMs), present an unprecedented opportunity to address signal loss and offer new, effective solutions for brands and agencies to connect with their ideal audiences without relying solely on user identification. We’re seeing major industry players signaling a move toward prediction over precision, relying much more on probabilistic data than in the past (source).
Coupling AI with extensive interest and contextual data can be particularly beneficial. A significant advancement in this space involves harnessing LLM-driven predictive methods. In traditional probabilistic modeling, algorithms analyze diverse signals such as user activity and interest data (collected in strict conformity with privacy regulations, leveraging the IAB Tech Lab’s GPP and TCF 2.2), alongside contextual signals – granularly categorized via the IAB Tech Lab’s Content Taxonomy 3.0. LLMs operate at an unprecedented scale and capacity, having been trained on extensive datasets and adept at comprehending human language and predicting missing words. This improves the precision of predictions derived from probabilistic data modeling, marking a new phase in predicting audience characteristics and enabling effective, large-scale targeting without relying on cookies.
Further innovations we’re excited about include AI-generated audiences, created by an AI translator tool. Essentially, the tool uses the text input from advertisers to create audience groupings that match the advertiser-specified requirements. AI can fully read and understand the description provided by brands and agencies and as such, generate specific audience segments – relying on context and predictions instead of cookies – that align with the provided criteria.
For example, an advertiser inputs “Adults up to 50 in Germany, who are price-sensitive, and interested in mobile phone contracts” as their intended audience. Then, the translator tool – an LLM-driven prediction model – interprets the prompt and translates it into a set of actionable signals. These signals could look like: Location: DE, Age: 18-50, Bargain Shopper, Phone model > 5 years old. The output is an AI-Generated audience segment, created as a weighted combination of contextual signals, interest data and predicted attributes.
AI can also boost the power of existing solutions like lookalike audiences and audience extensions. Lookalikes are audience segments created based on the similarities to an advertiser’s existing or seed audience. AI analyzes the characteristics and behaviors of the initial audience and then identifies and expands it to include individuals who exhibit similar attributes, and interests. This process extends the reach of the advertiser’s target audience without relying solely on the initial audience data, providing a broader scope for engagement. Current advancements in AI hold promise in significantly enhancing the effectiveness of lookalike audience modeling by enabling more intricate analysis and much more nuanced understanding of data, with real-time adaptation. Advanced AI models can adjust to changing trends and behaviors, continuously updating and refining lookalike audience profiles based on the latest data. This ensures audiences remain relevant and effective in dynamic environments, key in today’s world.
Conclusion
Our industry is going through a period of immense change – redefined by privacy and prediction – that requires adaptation, innovation, and agility. AI-driven targeting, reliant on privacy-safe data signals and granular contextual categories (for instance derived from the IAB Tech Lab’s Content Taxonomy 3.0), empowers brands and agencies to embrace this change, offering a clear path forward in an evolving landscape. By understanding the nuances of this new era and leveraging cutting-edge technology, they can not only navigate these shifts but also emerge as leaders.
VP of Product
Outbrain