The sentence that best describes e-commerce over the next several years was not spoken explicitly at Google Marketing Live 2026 — but it was present in every announcement made that day: the purchase increasingly does not happen on your website. With the launch of Universal Cart, Agent Payments Protocol, Universal Commerce Protocol, and the native integration of Klarna and Affirm, Google has built the technical infrastructure that allows AI agents to complete transactions autonomously on behalf of users. A user searches for a product. Gemini generates a conversational answer. Your product appears embedded in that answer. The purchase happens — without your website ever being visited.
Why This Is More Than a New Shopping Feature
Most brands have invested substantially in their website experience over recent years: checkout optimisation, landing page speed, conversion rate optimisation, UX testing. These investments retain their value — but they are no longer competing on the decisive battlefield. The decisive battlefield has moved upstream in the customer journey, to the moment when Google's AI decides which brand to recommend in a conversation. And that decision is not made on the basis of a landing page — it is made on the basis of product data quality, brand signal strength, and structured content attributes.
In concrete terms: Merchant Center data is no longer merely a channel asset — it is your storefront in AI commerce. Incomplete product feeds, missing attributes, or generic product descriptions cost visibility not only in Shopping tabs but across every conversational surface Google operates going forward. Brands with rich, structured, conversationally phrased product data get embedded. Generic brands get filtered out.
What This Means for Measurement
The quiet consequence of agentic commerce is a measurement gap. As conversions increasingly occur within the Google ecosystem — in AI Mode, via Universal Cart, through Agent Payments — they will no longer be fully visible in your analytics system. Last-click attribution already provides an incomplete picture of the customer journey; in an environment where the final touchpoint is literally outside your website, that incompleteness grows substantially larger.
Marketing Mix Modelling and incrementality testing do not become more optional in this environment — they become more essential. The brands that will make sound budget decisions in 2026 and beyond are those that have built their measurement infrastructure independently of the platform they are trying to measure.
Strategic Recommendations for Brands
Preparation for agentic commerce concentrates on three areas of action. First: invest in product data quality — complete attributes, conversational descriptions, high-quality imagery. Second: strengthen brand signals — reviews, authority indicators, trust factors — because these serve as the tiebreaker when an agent selects between comparable products. Third: build platform-independent measurement capable of giving you a complete picture of your marketing performance even when conversions occur outside your website's tracking environment.