The word “storefront” describes ecommerce from the perspective of a human visitor. It is the visible surface: navigation, collections, product pages, content, cart, and checkout.
Agentic commerce makes another layer equally important. Behind the storefront is a commercial information system that machines can evaluate without experiencing the page exactly as a person does. A useful name for this layer is the datafront.
The datafront is not a separate website. It is the organized commercial truth beneath the website: products, variants, attributes, relationships, inventory, policies, evidence, markets, and actions expressed coherently enough to serve multiple interfaces.
The storefront is one presentation of commerce
When the web browser was the dominant commerce interface, it was natural to design the business around pages. Product information was written for the PDP. Collection logic was designed for the category page. Policies lived in footer links. Recommendations appeared in visual carousels.
Now the same commercial system may be interpreted through search, conversational assistants, shopping agents, social surfaces, marketplace integrations, feeds, applications, and the merchant's own storefront.
The architecture therefore needs to separate commercial meaning from any one presentation.
What belongs in the datafront?
A Shopify datafront can include:
- products and variants as core commerce entities;
- Shopify product categories and merchant taxonomies;
- metafields for category-specific attributes;
- metaobjects for reusable structured entities;
- collections and explicit product relationships;
- inventory and availability states;
- market and localization context;
- pricing and purchase-option information;
- shipping, returns, warranty, privacy, and service policies;
- reviews, case studies, documentation, and other evidence;
- cart, checkout, support, quote, or other valid actions.
The visible storefront should render from this truth. Machine-facing channels should interpret the same truth. The closer those layers are aligned, the lower the risk that different interfaces describe the business differently.
Why page-first architecture creates problems
Page-first ecommerce often stores important facts inside presentation components. A compatibility rule might exist only in a paragraph. A service limitation might appear only in an accordion. An attribute could be embedded in an infographic. An important relationship might exist only because two products happen to be placed next to each other.
That works until another interface needs the same meaning.
The datafront approach asks a different question: What is the underlying commercial fact, and where should that fact live so every interface can use it?
Shopify is well suited to this architecture
Shopify's product model already separates products, variants, inventory, markets, and checkout from the visual theme. Metafields and metaobjects provide additional ways to model merchant-specific information. Themes then become one consumer of that structured system rather than the sole location of meaning.
This does not require every merchant to become a database architect. It requires discipline about what is a reusable fact versus what is presentation copy.
A product page becomes a view, not the database
Consider a technical product with compatibility requirements. In a page-first model, the merchant may write a long section listing compatible devices. In a datafront model, compatibility is represented structurally and the product page renders that information for the human visitor.
The advantage is not merely cleaner development. Other systems can now understand the relationship more reliably. The merchant updates the commercial truth once and multiple surfaces can remain aligned.
The datafront improves human UX too
Machine-readability and human usability are not opposing goals. Better structured data can power more accurate filters, comparison tools, specification tables, contextual FAQs, recommendations, reusable content, and consistent policy messaging.
In other words, the architecture built for MX can improve UX because both experiences draw from the same source of truth.
Think entity-first, relationship-first, state-first
A practical datafront design process asks three questions:
- Entities: what are the important things in this business—products, variants, services, experts, certifications, locations, specifications?
- Relationships: how are those things connected—compatible with, part of, substitute for, requires, available in, supported by?
- States: what can change—price, inventory, availability, eligibility, delivery, publication, subscription, market context?
Pages then communicate those entities, relationships, and states to people. Agents can reason over the same commercial system.
The future storefront is interface-agnostic underneath
Merchants do not know every interface through which customers will discover and buy in the future. That is precisely why the commercial model should not depend entirely on the current interface.
A strong datafront makes the store more portable across channels. It allows brand presentation to evolve without losing the underlying meaning. It also gives AI systems a cleaner representation of the business.
This is the architectural consequence of Machine Experience: design the commerce system first, then design its human and machine interfaces.
The previous article, The Machine Customer, explains why machines can enter before the storefront visit. Next, we examine one of the most important consequences of the datafront model: why product data can become more consequential than product page decoration.
