Marketers Are Drowning in Choices, Noise and AI Agents: Criteo’s Ed Dinichert

Why It Matters

The insights expose why marketers must simplify tech stacks and adopt realistic measurement, while Criteo’s data and self‑service tools position it to capture budget share from fragmented ad spend.

Key Takeaways

• Marketers face fragmentation across channels, platforms, and retail media
• Criteo’s shopper graph tracks 750 M users and $3 B daily transactions
• Incrementality remains unsolved; brands must craft pragmatic measurement frameworks
• Criteo Go offers AI‑driven self‑service onboarding for small merchants
• Model Context Protocols enable LLMs to execute campaigns via Criteo APIs

Pulse Analysis

The advertising landscape is at a tipping point, with marketers drowning in a sea of channels, measurement frameworks and AI promises. This fragmentation forces agencies and brands to spread budgets thinly, often without clear insight into which tactics truly move the needle. As the industry grapples with endless AI‑generated hype, the real challenge is cutting through the noise to focus on trustworthy data and incremental gains.

Criteo’s response leverages its deep commerce data moat—a shopper graph that reaches three‑quarters of a billion users daily and processes billions in transactions. By positioning this asset as the foundation for performance, Dinichert signals that the company is shifting from a legacy performance‑marketing label to a data‑centric platform. His candid admission that true incrementality remains elusive pushes marketers to adopt custom, pragmatic measurement models rather than waiting for a silver‑bullet solution.

The launch of Criteo Go marks a strategic pivot toward simplicity and scalability. The self‑service platform automates catalog ingestion, tag deployment and campaign setup using AI, lowering entry barriers for smaller merchants. Coupled with Model Context Protocols that let large language models interact directly with Criteo’s APIs, the company demonstrates a tangible, production‑grade AI use case. Together, these moves could reshape how advertisers allocate spend, favoring platforms that blend robust data, easy onboarding and realistic performance metrics.

Original Description

MIAMI – At the POSSIBLE conference, marketers arrived looking for clarity. Instead, they found another 47 measurement frameworks, 112 AI startups and at least three people claiming they had finally solved incrementality.

For Ed Dinichert, chief customer officer at Criteo, that chaos is precisely the point.

In an interview with Beet.TV contributor David Kaplan, Dinichert described an advertising industry struggling with fragmentation across channels, measurement systems and an increasingly crowded marketplace of AI-powered promises. His prescription was surprisingly simple: stop looking for magic bullets and start climbing the staircase one step at a time.

The identity crisis nobody expected

Having joined Criteo about five months ago, Dinichert said his biggest surprise wasn’t the technology itself but how differently the company is perceived compared with what it actually does.

"The biggest surprise has been really the difference of perception between what Criteo is perceived and what it is really about," Dinichert said.

While many advertisers still associate Criteo with its historical roots in performance marketing, Dinichert emphasized the scale of the company's commerce data assets. He pointed to a shopper graph that reaches 750 million users globally each day, maps 5 billion product SKUs and processes roughly $3 billion in daily transactions.

Performance, he said, has been a company obsession for two decades.

"There's a thousand engineers at Criteo that have been improving the performance algorithm to deliver performance," Dinichert said. "Performance has been an obsession in this company."

Welcome to the fragmentation Olympics

Asked about the biggest challenge facing marketers, Dinichert didn’t hesitate.

"It is definitely fragmentation," he said.

The problem extends from small businesses to large enterprises. Marketers face an expanding list of channels, platforms and retail media networks, all promising superior results. For smaller businesses operating with tighter budgets, choosing where to place their bets can feel less like media planning and more like a high-stakes game show.

Adding to the challenge is what Dinichert called "noise."

"Everybody's AI right now," he said.

That observation may qualify as the least controversial statement uttered at POSSIBLE all week.

Dinichert advised marketers to focus on trust and testing rather than getting swept up by every new AI-powered pitch deck promising autonomous campaign management, infinite optimization and perhaps world peace.

Incrementality remains advertising's unicorn

If fragmentation is the industry's headache, measurement may be its migraine.

Dinichert said one of the dominant themes he heard throughout POSSIBLE was marketers' search for incrementality.

"What I heard the most in the last 36 hours is everyone is searching for incrementality," he said.

His view was refreshingly realistic.

"If someone cracked incrementality in the last 20 years, it would be a mainstream model. And it's not," Dinichert said.

Rather than pursuing a perfect solution, he argued marketers should define their own measurement frameworks and identify the closest practical approximation to incrementality for their business.

In other words, advertising may never discover the Holy Grail. The best available option may be a very good metal detector.

When AI actually does something useful

Despite his skepticism toward AI hype, Dinichert described one emerging application that caught his attention.

He pointed to the growing use of Model Context Protocols, or MCPs, which allow AI systems to interact with software platforms and business tools.

As an example, he cited a recent project involving Dentsu in France, where an LLM was able to connect with Criteo's systems and facilitate campaign execution through existing APIs. The effort demonstrated a practical use case beyond the conference-stage promises currently flooding the market.

"That's real," Dinichert said. "That's a company, like, they delivered."

He expects adoption to take time, noting that some of the most innovative implementations are coming not from massive engineering organizations but from smaller teams willing to experiment.

Criteo bets on simplicity

A major part of Criteo's current strategy centers on Criteo Go, a self-service platform launched in April.

According to Dinichert, it represents the company's first intentionally designed self-service user interface in its 20-year history. Early feedback has focused on usability and onboarding, particularly for smaller merchants.

The platform includes an AI-driven onboarding process that can automatically ingest product catalogs from Shopify stores, configure tracking tags and prepare campaigns with minimal manual effort.

Dinichert also said the company's agentic technology has produced performance improvements that exceed results from some managed-service campaigns.

"We have enough uplift of this agentic technology being better than our