Agentic Commerce Readiness: A 2026 Checklist for Stores

Agentic commerce readiness is the new table stakes for online stores: AI shopping agents are placing real orders today, and most catalogs are simply invisible to them. When a buyer asks ChatGPT, Perplexity, or Google AI Mode to “find and buy a waterproof duffel under $120 with free returns,” the agent only shortlists merchants whose data it can actually read, qualify, and check out from.

What changed is the economics. Adobe Analytics found AI-sourced traffic to U.S. retail sites grew 393% year over year in the first quarter of 2026, and in March 2026 that traffic converted 42% better than organic — a complete reversal from a year earlier, when AI visitors converted worse. The window between “agents are coming” and “agents are buying” has effectively closed.

This guide is a merchant self-audit, not another “agentic commerce is the future” explainer. It walks six readiness pillars — product feed, structured data, protocol integration, machine-readable policies, trust signals, and AI citation — and gives you a 0–18 scoring tier so you can see exactly where an agent stalls when it tries to buy from you.

For most of 2025, merchants treated AI traffic as low-quality curiosity clicks. Adobe’s own data backed that up: in March 2025, AI-referred visitors to U.S. retail sites converted roughly 38% worse than organic. Twelve months later the picture inverted. By March 2026, AI-referred traffic converted 42% better than organic — a record high and a swing of about 80 points in a single year, measured across Adobe’s data set of more than a trillion visits.

The behavioral signals explain why. Adobe reported AI-referred visitors spent 48% longer per session and viewed 13% more pages per visit — they arrive pre-qualified by the agent that sent them, having already narrowed the choice set. The interesting read here is that agentic traffic is not a new acquisition channel so much as a higher-intent filter on demand you were already competing for. The merchants who win it are the ones an agent can confidently shortlist.

Forecasts about how far this goes vary wildly and should be read with caution — a widely cited Gartner projection has agents mediating a sizable share of online retail within a couple of years, but the primary report is paywalled and hard to verify, so treat such headline numbers as directional rather than gospel. The defensible point is narrower and more useful: agent-led demand is already converting better than the channels you optimize for today, and the cost of being unreadable is rising every quarter.

The hard part of agentic readiness is not a single integration — it is the fragmentation. Four standards now coexist, each backed by a different coalition. The Agentic Commerce Protocol (ACP) was announced by Stripe, OpenAI, and Meta on September 29, 2025 under an open Apache 2.0 license, and it powers ChatGPT Instant Checkout (live first on Etsy in the U.S., with Shopify’s 1M+ merchants rolling out). Google and Shopify’s Universal Commerce Protocol (UCP) added Cart, Catalog, and Identity Linking on March 19, 2026 and is now endorsed by 20+ organizations including Mastercard, Visa, and American Express. For the full technical walkthrough of ACP, see our Agentic Commerce Protocol deep dive, and for how the open standards line up against one another, the UCP vs ACP vs AP2 merchant guide.

Two more layers matter. Adyen Agentic, announced June 16, 2026, is the only stack that explicitly bridges ACP, UCP, and AP2 at once — though it is in limited availability for U.S. enterprise merchants only, not something an SMB can switch on today. And MCP (Model Context Protocol), the Anthropic standard Shopify has been rolling out to its stores, exposes your catalog for agent discovery and authenticated order access. The table below puts the four side by side.

A note of caution on AP2 (Agent Payments Protocol): it appears in Adyen’s and Google’s materials, but its public documentation is still thin, so we treat it as an emerging layer rather than a settled peer of ACP and UCP. The practical implication of the whole picture is a protocol-hedge problem: a store that integrates only ACP may be invisible to Gemini-sourced buyers, and a store on UCP alone may miss native ChatGPT checkout.

Readiness is not binary. A store can be perfectly discoverable yet impossible to transact with, or technically transactable but never recommended because its trust signals are missing. The framework below scores each of six pillars from 0 (not started) to 3 (optimized), for a maximum of 18 points. Add the pillar scores and map the total to a readiness tier.

Everything else depends on the data layer. An agent cannot recommend what it cannot parse, and it cannot qualify what it cannot read. Pillar 1 is your product feed: a complete, machine-readable catalog with real-time pricing and inventory. Shopify’s Catalog auto-enriches and syndicates merchant products into a universal taxonomy — Shopify reports that AI searches powered by its catalog convert at roughly twice the rate of those using scraped data, and that it has deployed MCP endpoints across millions of stores. Treat both as vendor-stated figures, but the direction is clear: agents reward clean, first-party feeds over scraped guesses.

Pillar 2 is structured data. JSON-LD Product markup is what lets an agent read your catalog without rendering your page. A practical floor is name, brand, description, image, price (with priceValidUntil), availability, and aggregateRating; a fuller implementation adds GTIN/MPN, review nodes, hasMerchantReturnPolicy, and shippingDetails. Third-party analysis suggests the majority of pages AI engines cite carry structured data and that comprehensive schema correlates with materially more frequent inclusion — directional, not audited, but consistent with how these systems work. Our companion guides cover getting your product data AI-ready and schema markup strategy after March 2026.

Discoverability gets you cited; Pillars 3 and 4 get you bought from. Pillar 3 is protocol integration — wiring your store into the checkout standards from Section 02 so an agent can complete a purchase, not just link out. Pillar 4 is machine-readable policies: returns, refunds, and shipping expressed in structured markup, not buried in a prose page. This matters because agents qualify merchants against a buyer’s constraints before recommending — “must allow returns within 30 days,” “must ship by Friday.” An agent that cannot parse your return window cannot promise it to the shopper, so it routes around you.

The last two pillars are where readiness becomes preference. Pillar 5 is trust signals and agent permissions. Checkout.com’s June 2026 study found consumers want concrete controls before they delegate a purchase: spending caps, instant revocation, and easy cancellation top the list, and 75% of merchants say real-time permission revocation will be critical for adoption. The average spend a consumer will let an agent commit without approval was about £177 — useful as a default ceiling when you design confirmation loops. Our analysis of the trust gap between what consumers expect and what merchants have built goes deeper on this consumer data.

Pillar 6 is AI citation optimization — being the source an agent quotes and links. Enroll in merchant programs (the Perplexity Merchant Program is free and takes minutes; products not in it simply do not appear in Perplexity shopping results). Keep product content fresh, factual, and entity-linked rather than marketing fluff; SEO monitoring tools observe that recently updated pages appear in AI answers more often, and that ChatGPT drives the bulk of AI referral traffic while Perplexity contributes a smaller, high-intent slice — directional observations, not audited shares. Conversational and voice agents share this discovery layer, so the same discipline that wins conversational shopping optimization compounds here.

You do not need all six pillars at once. Score yourself, then start with the cheapest points that move you up a tier. For almost every store, that means the free, channel-agnostic pillars first — schema, machine-readable policies, and the Perplexity Merchant Program — before any protocol integration work. From there, the path forks by platform.

However you sequence it, the deeper Shopify-side detail lives in our Shopify Spring 2026 agentic commerce rollout companion. If you would rather not assemble the stack yourself, our ecommerce engagements run exactly this audit — feed, schema, protocols, policies, trust, and citation — and our agentic SEO work handles the discovery and citation layer so agents find you in the first place.