Criteo, the advertising technology company, has unveiled its latest foray into artificial intelligence with new integrations and pilot programs, marking a significant step in its evolution within the AI landscape. The announcement came on Tuesday, shortly after the holiday shopping season, as Criteo aims to enhance its offerings in the competitive ad tech market.
At a recent event for reporters, Criteo introduced its integration with a large language model (LLM) through its “Model Context Protocol” (MCP). This framework is designed to allow LLMs to tap into Criteo’s extensive shopper data. According to Criteo CEO Michael Komasinski, the integration aims to determine if utilizing Criteo’s data can yield improved results for shopping-related prompts. Notably, the end user of the LLM will not need to be familiar with Criteo, as the company’s product recommendation system will underpin the LLM’s suggestions.
However, the specifics of how this partnership might operate in practice remain unclear. Komasinski acknowledged that revenue generation strategies have yet to be established, with potential avenues including licensing arrangements similar to those seen in media industries, or more traditional advertising models based on performance metrics. “I think it gets a lot more interesting in the context of a native ad solution,” he remarked, while also noting the possibility of charging fees based on LLM usage metrics, such as per crawl or query.
As Criteo ventures deeper into AI, it faces the challenge of solidifying its position as a credible data provider in this rapidly evolving field. Demonstrating the effectiveness of its data as a training resource for LLMs could position the company favorably within the industry. Komasinski emphasized the importance of data sets like Criteo’s, which can significantly enhance LLMs’ capabilities in commerce applications. With approximately 720 million daily active users, Criteo is poised to leverage its extensive data resources, which include insights from $1 trillion in observed gross merchandise sales last year.
Despite these promising figures, it is essential to approach them with caution. Komasinski’s reference to Criteo’s DAUs pertains to the number of individuals targeted through its online advertising, rather than users directly interacting with its platform. Moreover, while Criteo claims to observe vast transactions, it does not own the underlying data, which presents limitations in its scope of operation. In contrast, major competitors like Amazon and Shopify have restrictive policies regarding LLM access, limiting potential partnerships that could enhance AI services.
In addition to its LLM activities, Criteo is launching a product called Audience Agent, designed to simplify campaign setup and analytics for users without advanced data expertise. This new tool enables users to create advertising campaigns through natural language prompts, streamlining the process to a six-click system—from audience generation to campaign activation. However, the effectiveness of Audience Agent remains to be seen, as current audience segment suggestions appear somewhat predictable, prompting questions about the tool’s ability to uncover more nuanced marketing insights.
According to Criteo’s executives, advertisers are increasingly focused on incremental return on ad spend (ROAS), thus shifting the company’s strategy toward upper-funnel advertising solutions. This change is evident in Criteo’s recent expansion into platforms such as Meta and TikTok, which cater to consumer engagement and shopping considerations. The company must now validate its advertising expenditures on these channels, even as brands manage separate social media campaigns.
Criteo’s ambition to broaden its reach in the advertising space resonates with the industry’s shift towards integrated digital marketing strategies. As the company continues to explore AI’s potential applications, its success in proving its capabilities will be critical in establishing itself as a viable alternative to tech giants like Google and Meta. “A point which I think about a lot,” Parsons said, “is how far can we go into that upper funnel with those unexpected correlations and still prove that we’re a reasonably good alternative to Google and Meta? Time will tell.”
See also
Langsmith Launches No-Code Agent Builder for Custom Market Research Assistants
AI Transforms Advertising: Market to Reach $107.5 Billion by 2028 with Enhanced Targeting
BrandMeOn Launches AI Chatbot for Instant Customer Engagement on Social Media Platforms
SharpenCX and Ytel Merge to Launch Unified AI CX Platform for Seamless Engagement
Retail AI Council Launches Shopper Context Protocol to Enhance Customer Loyalty Amid AI Shift
