The most effective way to prepare your product pages for agentic commerce is to stop thinking about them as pages at all. You need to start treating them as structured, queryable data assets. For years, we’ve obsessed over the visual presentation of a PDP, focusing on button colors, image placement, and scroll depth. These things still matter for the human-facing web, but the rise of AI shopping agents means your primary "visitor" is increasingly a robot, and this robot doesn’t care about your beautiful lifestyle photography. It cares about clean, comprehensive, and well-organized data it can parse to make a decision on a user’s behalf. Success in this new era isn’t about radically new technologies, but a radical commitment to the fundamentals of product information.
This shift is a direct response to how AI assistants are changing discovery and purchasing. On their show Ecommerce Playbook, Richard and Tony make the point that well-structured product data feeds are becoming the absolute bedrock of modern commerce. When a user asks an AI, “Find me a waterproof, breathable, lightweight running jacket under $200 with reflective details,” the AI isn’t browsing your website. It’s querying a protocol or a data feed that contains the attributes of your products. If your jacket’s data doesn’t explicitly state "breathable" or "reflective details" in a machine-readable format, your product is simply invisible. Your PDP has become an API endpoint for AI, and if the data is messy, incomplete, or purely locked in visual formats, you don’t exist in this new channel.
The New Foundation: Structured Data and Persuasive Content
This is where the tactical work begins, and it’s less glamorous than most people hope. The job is to translate every compelling aspect of your product into structured data. On the Shopify1Percent podcast, Gavin McKew lays out a brilliant, actionable framework for this. He advises starting with your top-selling SKUs and methodically enriching them with Shopify Metafields. This isn’t just about basic specs like size and color. It’s about creating fields for your product’s core purpose, the problem it solves, the ideal customer, materials, key features, and unique benefits. This is how your brand’s voice and perspective survive in a world of aggregated, agent-driven results.
Think of it this way: your marketing copy tells a story, and your metafields must provide the structured outline of that same story. If you sell a skincare product, your data should include not just the ingredient list, but also fields for skinconcern (e.g., "acne," "dryness"), keybenefit (e.g., "reduces redness," "hydrates for 24 hours"), and usage_instructions. This level of detail is what allows an AI to confidently match your product to a user’s complex query. This is a core theme in the broader discussion around AI in e-commerce; the technology is only as good as the data it’s given. This isn’t just about being found; it’s about being chosen.
Where Conventional Wisdom Goes Wrong
Where many operators get this wrong is by treating it as a one-time technical SEO project. It’s a continuous merchandising and brand function. The data needs to be as compelling and persuasive as your ad copy. Furthermore, as Tim Peter points out on Thinks Out Loud, there’s a significant gap between the grand vision of agentic commerce and its current, practical reality. Agents are still being developed, and their decision-making processes are evolving. This means we are in a transitional phase. Human shoppers will still be visiting your PDPs directly for the foreseeable future, so you can’t abandon the human-centric experience. You must maintain two layers: a visually compelling and intuitive front-end for people, and a rigorously structured and detailed back-end for machines.
Another misstep is ignoring the ecosystem. An agent won’t just look at your data; it will look for social proof, price consistency, and availability across the web. Gavin McKew stresses the need to tighten price parity across all channels. If an agent sees your product for $100 on your site but $90 on Amazon, it creates distrust and friction, potentially disqualifying your direct offer. Your product data, pricing, and inventory levels must be a single source of truth everywhere your brand appears.
Your First 90 Days
This can feel overwhelming, so I’d follow a structured plan. Don’t start by building; start by auditing.
Days 1-30: Audit and Strategize. Take Gavin McKew’s advice from Shopify1Percent and use ChatGPT as a shopper for your own products. Ask it questions you know your customers ask. "I need a carry-on bag that fits under the seat on Spirit Airlines." "Find me a non-toxic pan that can go from the stovetop to the oven." Log every single gap where the AI fails to find or properly understand your product. This audit becomes the blueprint for your metafield strategy. Identify the 10-15 most important data points that are currently missing from your top 20% of SKUs.
Days 31-60: Build and Enrich. This is the heads-down work. Based on your audit, build out the necessary metafields in Shopify. Then, begin the manual (or semi-automated) process of populating this data for your top-selling products. This is the moment to translate your brand’s soul into structured data. What problem does this product really solve? Who is it for? Why is it better than the competition? Put that in the data. This is the first step toward building out your data-driven decision-making capabilities for this new channel.
Days 61-90: Expand and Refine. Now, expand your enrichment process to your long-tail SKUs. Simultaneously, work with your team to ensure absolute price and inventory parity across every channel you sell on. As part of this, you can start exploring more advanced tactics, like creating systems to pipe structured data from customer reviews (e.g., common themes, pros, and cons) into your product records. The world of contextual commerce, which the team at Future Commerce has covered extensively, is one where the agent provides the context and your PDP provides the perfect, data-rich answer. Preparing for that future starts today, with a spreadsheet or a metafield manager, and a deep understanding of what your product truly is.
