Agentic Checkout Is Live. Your SKU Data Isn't Ready.

May 19, 2026. Google I/O. Universal Cart goes public. One day later, Klarna's Shopping Search app is live inside ChatGPT. Two separate announcements. One signal: the checkout event is no longer happening on your storefront. It's happening inside an agent. If your product data isn't structured for machine consumption right now, you are invisible to both surfaces.

What the Average Brand Gets Wrong About This

Most operators are treating agentic commerce as a marketing problem. It isn't. It's a data infrastructure problem. Google's Universal Cart pulls product attributes, pricing, and availability through structured feeds. Klarna's app queries purchase intent through conversational AI, then surfaces SKUs that match. Neither surface browses your homepage. Neither reads your brand story. They read your data schema.

The average brand has product titles written for human keyword scanning. Attributes are inconsistent across catalog. Bundle logic isn't machine-readable. Top-decile operators—the ones already running clean SP-API integrations and structured Google Shopping feeds—are positioned to have their ASINs and product nodes resolved by these agents automatically. The gap between average and top decile isn't ad spend. It's data hygiene.

The Benchmark Gap in Three Tiers

Tier one: most brands. Product feeds updated manually, quarterly. Attribute completeness under 70%. No schema markup beyond basic price and availability. These brands will not appear in Universal Cart queries. They will not surface in Klarna's AI search. Tier two: the top 25%. Automated feed management through a third-party tool. Attribute completeness above 85%. Some schema markup in place. These brands will surface, inconsistently. Tier three: the top decile. Full schema markup. Automated feed sync tied to inventory velocity. Attribute fields complete including use-case, compatibility, and material specs. These brands get resolved first by agents optimizing for purchase completion. That resolution happens before a human ever sees the result.

Home Depot Shows You the B2B Version of This

Home Depot's Q1 move is worth reading as a parallel case. They consolidated AI tools for Pro customers into a single workspace. That decision is structurally identical to what top-decile DTC brands need to do with their product data layer. Fragmented data across multiple tools produces fragmented agent outputs. One clean source of truth—product attributes, pricing logic, inventory status—produces consistent resolution across every surface. Home Depot is solving this for a B2B buyer cohort. Your brand needs to solve it for every agent that might place an order on a consumer's behalf.

Three Actions. Execute in Order.

First: audit your product feed attribute completeness today. Pull your catalog. Count the fields. Any SKU below 90% attribute completeness is a SKU that won't resolve cleanly in an agentic query. Fix those before you touch anything else. Second: implement structured data schema markup across your PDPs. At minimum: Product, Offer, AggregateRating, and ItemAvailability. These are the fields Google's Universal Cart reads. These are the fields Klarna's app parses. Third: connect your inventory system to your feed in real time. An agent that surfaces a product with stale availability data and the order fails—your brand gets a negative signal in that agent's resolution logic. Sell-through accuracy and feed sync frequency are now ranking factors.

Three Questions to Pressure-Test Your Readiness

Can you state, right now, what percentage of your SKUs have complete attribute data across every selling surface? If your feed breaks tomorrow, how many hours before an agent surfacing your products returns a null result? For your top 20 revenue ASINs—does your structured schema markup reflect current pricing, availability, and product specs, or the version someone uploaded eight months ago? Answer those three honestly. That answer is your starting point. Pull the feed report now.