An Overview of Agentic Commerce—and How It Impacts Media Planning
Management Summary
There are moments when you realize that something is changing fundamentally. For me, Agentic Commerce is one of those moments. We’ve been exploring it intensively for months: in client meetings, on panels, and most recently at the IAB Network Event. And the deeper you dive into it, the clearer it becomes that we’re currently standing at a turning point in media planning.
The Shift
We’re familiar with traditional e-commerce: Someone searches for a product, lands on a product page, reads through all the details, makes a decision based on that information, and makes a purchase. Every step of this journey presented an opportunity for advertising—from the awareness campaign to search ads, optimized landing pages, and retargeting.
Agentic Commerce disrupts this process. The new workflow: Someone tells their AI assistant what they need, and the agent compares, filters, decides, and makes the purchase. Directly, without a product page, without a hero image, and without a brand story at the point of sale. What ultimately remains is data: the product feed with all the important attributes and signals.
Google now has over 20 global partners on board for the Universal Commerce Protocol—from Shopify to Zalando to Mastercard. Walmart can be booked through Gemini. OpenAI has rolled out “Buy it in ChatGPT,” which will soon be available to one million Shopify merchants. And at Google Marketing Live in May, the next building blocks were laid with the Universal Cart and the Agent Payments Protocol.
The technology is here. It continues to evolve. For the DACH market, we expect the first launches in 6 to 12 months, with widespread availability by 2027. So we have some lead time, but we need to make the most of it. Preparing an organization for Agentic Commerce doesn’t happen overnight.
Advertising for people. And advertising for their agents.
This is where the real challenge for media planning lies. We’ve always had a clear target audience: people. Media planning meant: Who should see my message, when, where, and how often? All models, all KPIs, and all booking logic are based on this fundamental assumption.
Now there are two recipients: humans and their agents. And the two operate according to fundamentally different logics.
- People respond to emotion, relevance, aesthetics, and trust. They are influenced by a powerful campaign, a good story, and a brand image that has been built up over the years.
- The agent responds to data quality, completeness, and structure. It does not read continuous text or evaluate visual elements. It compares attributes—such as price, availability, return rate, or category. If the feed is not fully maintained, the product simply does not exist for it. Regardless of how strong the brand behind it is.
But the feed is just one source. What agents also evaluate is what the internet says about a brand: reviews, mentions, discussions, and user-generated content. A product that real people talk about positively—in reviews, forums, or on social media—sends signals of relevance. The fact that all these signals are also linked to branding is often overlooked. But getting people to talk about your brand feeds precisely the sources from which agents draw their evaluations. Online visibility has always been valuable—in the age of agents, it’s becoming an even more important foundation for purchasing decisions. [This is where Generative Engine Optimization (GEO) comes into play: This new discipline ensures that brands and products are featured as reliable sources in the AI-generated responses from chatbots and assistants.]
The ability to leverage both is the new core competency. And most media plans today still address only the first:
Traditional media buying is structured around a funnel that runs through a company’s own channels: awareness on high-reach platforms, performance in the middle and lower parts of the funnel, and conversions on the company’s own website. What happens to this model when the funnel no longer runs through the company’s own website?
Brand investments haven’t lost their relevance—quite the opposite. But their purpose is shifting. When someone tells an agent, “Buy me running shoes from a sustainable, European brand,” they’ve picked up that preference from somewhere. Brand communication is responsible for that. It’s just that the impact no longer comes from clicking on an ad—it comes from the instruction someone gives their agent.
So we need to learn to justify brand investments using a new logic—not based on click-through rates, but on shaping consumer preferences. That’s harder to measure, but it’s what really matters. And at the same time, entirely new advertising formats are emerging: Conversational Discovery Ads, Highlighted Answers in AI Mode, or recommendations within Gemini conversations. No traditional media plan today is geared toward these.
When do we need to take action?
Now. Not because Agentic Commerce will take over all e-commerce tomorrow. But because preparation takes time. Feed quality is already paying off in traditional shopping. Conversational Attributes—the new, expanded product descriptions that Google is currently piloting—aren’t just a feature upgrade. They’re the content that the agent reads. So if you don’t fill them out today, you’ll be invisible tomorrow.
On top of that, the organizational issue is often the most difficult. Who in the company is responsible for AI search visibility? Who decides whether an agent is allowed to sell on our behalf?
Conclusion
The role of media planning is shifting. It’s no longer just about allocating budgets across channels. It’s about generating visibility—through both human touchpoints and automated decision-making processes at the same time.
Specifically, this means: feed optimization for agents. GEO as a discipline alongside SEO. Measurement concepts that track AI search visibility. And an understanding of why investing in data quality is also a media decision.
Agentic Commerce is a shift that is currently taking place—at different paces, depending on the market. The question is no longer “if.” The question is, who will be ready when it fully takes hold?
