Shopify published an executive guide to agentic commerce last week that’s worth your time if you run an e-commerce brand. I’m going to save you the read and pull out the parts that matter most — the practical stuff you can actually do, not the parts that are selling you on Shopify.

The core question they pose is a good one: are your products actually legible to the agents that are deciding what to recommend? Most brand leaders think their product data is AI-ready. What they actually have is product pages built for humans to browse. Those are two very different things.

Here’s the line that should stick with you: “Unlike a low-ranking Google result, there’s no page two to scroll to.” When an agent doesn’t understand your product, it doesn’t rank you lower. It skips you entirely and recommends a competitor. There’s no second chance in a conversation.

The guide lays out a three-phase approach. I’m going to compress it into ten things you can do this week — because most of Phase 1 is just checking boxes that should’ve been checked already.

  1. Audit your product data like an agent would read it. Not your product pages — your actual data. Are your attributes structured? Do you have materials, dimensions, compatibility, fit, and use cases in discrete fields, or is all of that buried in a paragraph of marketing copy? Agents parse structured fields. They skip prose. A description that says “our most comfortable shoe yet, designed for the modern explorer” tells an agent nothing. “Men’s trail running shoe, 10.2 oz, 6mm drop, Gore-Tex upper, Vibram outsole” tells it everything.
  2. Make shipping, returns, and warranty machine-readable. The Shopify guide calls this out specifically — if your return policy lives in a PDF or an FAQ page with no structured markup, agents can’t compare you against competitors on fulfillment terms. Shipping speeds, return windows, warranty details need to be structured at the product level, not the site level. Boring work. Almost nobody has done it. That’s why it’s an advantage right now.
  3. Normalize your images. Consistent angles, proper alt text, standardized naming conventions. Agents are starting to pull image data as a quality signal. If your product photos are a mix of lifestyle shots and poorly cropped manufacturer images with file names like IMG_4392.jpg, you’re losing structured credibility.
  4. Check your robots.txt. ChatGPT uses a crawler called OAI-SearchBot to index your site for shopping results. If it’s blocked, your products don’t exist in ChatGPT. Same goes for GPTBot, ClaudeBot, and PerplexityBot. Allow them on product and catalog pages, block checkout and account paths. Most brands haven’t touched their robots.txt since 2023. This takes five minutes and it’s the single biggest reason brands are invisible without knowing it.
  5. Get into Shopify Catalog. This is the Shopify-specific play, but it’s worth calling out because the distribution is real. Shopify Catalog feeds into ChatGPT, Microsoft Copilot, Google AI Mode, and the Shop App through a single integration. You don’t need to be on Shopify — the Agentic Plan lets any brand add products to the catalog. No transaction fees beyond standard processing. The point isn’t “use Shopify.” The point is that maintaining separate integrations for every AI surface doesn’t scale, and Shopify is currently the widest pipe.
  6. Treat GEO like you treat SEO — as a weekly practice. The Shopify guide introduces Generative Engine Optimization as a discipline, not a one-time project. Refresh your product detail pages, FAQs, and collection pages regularly for AI extraction clarity. And don’t ditch your SEO fundamentals — AI chatbots still use search engines as input. ChatGPT pulls 75% of its product data from Google Shopping. Your Google Merchant Center feed IS your ChatGPT strategy.
  7. Feed your brand context to agents directly. The guide mentions a Knowledge Base App that lets you give agents answers to edge cases — return policies for opened products, compatibility questions, sizing exceptions. The broader principle applies regardless of platform: if an agent can’t answer a follow-up question about your product, it loses confidence in recommending you. Think about the ten most common customer service questions for your top products and make those answers structured and accessible.
  8. Set up AI traffic attribution now. You need to know what’s coming from agent-driven sources and what isn’t. The Shopify guide recommends tracking AI-driven traffic, landing pages, query patterns, conversion rates, return rates, and support contacts. Measure AOV and return rates for agent-referred orders separately. Are they higher quality (Criteo says LLM-referred traffic converts 1.5x better) or are they price-hunting orders at your worst margin? The answer changes your strategy.
  9. Run a controlled test in one category. Pick one or two high-intent product categories. Enrich the attributes, add structured FAQs, tighten the imagery, clean up the policies. Measure the lift. The Shopify guide calls this a Phase 2 move, but you can start it this week with a small catalog. You’ll learn more from running one enriched category against your existing data than from reading another guide about agentic commerce.
  10. Assign someone to own this. The Shopify guide buries this in Phase 1 but it might be the most important item. Agentic commerce readiness touches product data, marketing, IT, and operations. Without a cross-functional owner, it’ll sit in the gap between teams. This doesn’t need to be a new hire. It needs to be someone’s explicit responsibility, with authority to push catalog changes, policy updates, and feed improvements through approval.

The Shopify guide frames this as a three-phase roadmap: get fundamentals agent-ready, make it discoverable and measurable, then scale. That’s the right sequence. But most brands are still at step zero — they haven’t even tested whether an agent can find their products. Open ChatGPT right now and ask it to recommend something in your category. If you’re not in the answer, that’s your starting point.