The news: Retail media is maturing faster than the systems used to measure it. That was a clear message from this month’s Commerce Media Brand Summit conference, where executives described an industry that has built out networks and ad formats but is still struggling to prove value with enough speed, consistency, and depth.

In recent years, the industry established retail media networks and expanded ad formats. But as growth eases—our forecast calls for global retail media ad spending to rise 15.4% this year and 10.2% in 2029—the focus is moving to accountability.

“Brands still need more data, more granularity,” Mert Damlapinar, founder of digital commerce insights and media platform eCommert, told EMARKETER. He said advertisers want better ways to evaluate performance, compare results across platforms, and turn campaign data into strategic decisions.

Rethinking metrics: Though return on ad spend (ROAS) remains a core efficiency measure, brands want to know whether their spending gained incremental sales, won new customers, or changed behavior.

One reason measurement remains a problem is the lack of consistency across providers. “Different methodologies to measure incremental return on ad spend makes it hard to truly compare apples to apples,” said Cristina Costa, senior vice president for sales at analytics startup Peq. Costa said her company uses machine learning to standardize measurement across marketing channels—including digital, print, in-store, sampling, and experiential—to support real-time forecasting and planning.

That’s where AI is starting to redefine the measurement conversation.

“AI is fundamentally changing what retail media measurement can do,” said Hemang Upadhyay, an AI researcher. “It’s moving the industry beyond basic attribution toward real diagnostic intelligence.”

Upadhyay said AI can analyze far more signals across channels to explain not just what converted, but why—including patterns in audience behavior, creative effectiveness, timing, and product-level demand. That, he said, can help brands and retailers forecast outcomes more accurately and identify incremental opportunities traditional reporting often misses.

But Upadhyay said the biggest constraint is not the sophistication of AI algorithms but rather the quality and consistency of the underlying data. Fragmented platforms, inconsistent taxonomies, and varying definitions of incrementality still limit how much AI can deliver.

Implications for brands: As retail media growth matures, brands are likely to face greater pressure to justify spending with more than surface-level ROAS. That will put a premium on cleaner data, stronger measures of true ad effects, and tools that can identify performance issues faster to support quicker spending decisions.