Publicis Groupe to buy Lotame in a rare instance of agency-led ad tech consolidation

The purchase also marks a significant expansion of its global identity and data capabilities, especially in APAC and Europe, where its new acquisition has a strong presence. Lotame was already Publicis Groupe’s data partner in both regions and helping to power CoreAI applications. Publicis also plans to expand its publisher network by integrating Lotame with Epsilon’s first-party tagging system to help publishers navigate cookie deprecation.

Located in more than 100 countries, Lotame adds another 1.6 billion global IDs to Publicis’ Epsilon’s platform, taking the total size of its audience graph close to 4 billion unique profiles. The exact number will be known when the deal closes in the second quarter, and the company spends a few months de-duplicating the datasets.

Lotame’s global data assets will feed into Epsilon’s “identity spine,” strengthening Publicis’ vertical connected identity solutions — a growing priority as brands seek more control over their data strategies. 

According to Publicis’s chief solutions architect, Scott Hagedorn, the holding company is evolving its identity and AI strategy by shifting from a centralized onboarding model to an “Identity-as-a-Service approach.”

Hagedorn said the holding company has already launched 50 pilots of CoreAI, a platform it debuted last year as part of a $300 million scheme. One example includes using probabilistic modeling to help a financial services client prospect consumers likely to become high net worth. This was done by using an existing Epsilon data with a two-year look-back window and combining it with another data set showing people who had just become high net.

“You can rewind the tape on those individuals and see what they looked like two years ago, what their attributes were, and then grab those attributes and pull them forward and model what people that look like those attributes now,” he said.

Improving probabilistic modeling is part of the plan for integrating Lotame within Epsilon’s identity spine. Another example of a pilot involves helping CPG advertisers assess the impact of inflation on consumer spending, such as whether shoppers will trade up or down between premium and value brands. 

It could also help expand retail media capabilities by linking local store-level data (e.g., stockouts, price changes) with consumer profiles and then making real-time adjustments to a brand’s media bidding strategies to attract the required audience.

To improve privacy controls, Publicis plans to deploy AI models within client-owned data environments rather than relying solely on centralized processing. It is also developing a “probabilistic data workbench” to provide more transparency for how AI-driven models generate projections and work with incomplete datasets.

To improve the accuracy of large language models and other machine learning tools, Publicis uses census-level data and industry-specific sources like Circana for CPG and Experian for automotive. It’s also in the process of rolling out natural language processing (NLP) capabilities into its media operating system so users can query data using conversational prompts instead of relying on static dashboards.