How Brands Are REALLY Using AI for Retail Media Automation
E154

How Brands Are REALLY Using AI for Retail Media Automation

At Xnurta’s Signal to Scale Summit this week, I heard some of the most practical conversations yet about how brands are actually using AI in retail media. Spark Foundry’s Kris McDermott summed up the mood perfectly: “We did AI. To what end? That is not a verb.” 

From Boiron managing 50,000 keywords with a two-person team, to AI-powered creative testing that drove a 29% sales lift, this episode explores where agentic workflows are delivering real operational value, and where they still require strong human oversight.

I also break down why advanced practitioners are moving beyond ROAS as the primary metric, what Amazon Ads is building with MCP servers and Skills, and why Xnurta's new open-source AI evaluation framework could become critical infrastructure for the future of retail media automation. The biggest takeaway: AI can accelerate performance, but only when paired with clear goals, guardrails, and teams that still understand “the gears” behind the machine.

This episode is sponsored by Mirakl Ads

Timeline

[00:00] — “We did AI. To what end?” and the real theme of the summit 
[01:18] — Boiron’s AI-powered retail media workflow managing 50,000 keywords 
[02:00] — AI image scoring drives a 29% sales increase 
[03:30] — Why ROAS alone is not the goal anymore 
[04:15] — The “automatic vs stick shift” analogy for AI media buying 
[06:04] — Xnurta launches open-source AI evaluation framework for retail media

Links & Resources