Shopify Storefront MCP vs Customer Accounts MCP: What's the Difference?

When Shopify introduced native Model Context Protocol support in 2026, it shipped two separate MCP server connections, not one. Each connection exposes a different category of store data to AI agents and serves a fundamentally different purpose in the customer journey.

Understanding the distinction matters because implementing the wrong connection first, or assuming one replaces the other, leads to an incomplete AI integration that either misses the pre-purchase discovery phase or leaves post-purchase customer service entirely unaddressed.

The Two Shopify MCP Connections at a Glance

Storefront MCP: Pre-Purchase AI Access

The Storefront MCP connection is the public-facing AI layer of your Shopify store. It requires no customer authentication, any AI agent that discovers your MCP endpoint can query your products, collections, and store information without a customer needing to log in.

This is the connection that makes your store discoverable and transactable by external AI assistants. When a customer asks Claude or Perplexity to recommend a product that your store carries, the Storefront MCP is what enables those agents to query your live catalogue, retrieve matching products, and generate a checkout link on the spot.

What the Storefront MCP Handles

• Product discovery, search by keyword, category, price range, tags, and metafield attributes
• Product detail, full specifications, variant data, metafields, images, pricing, and availability
• Collection browsing, category-level navigation with filtering
• Cart creation and management, programmatic cart building with checkout URL generation
• Store information, policies, contact details, shipping information
• Real-time inventory, live stock levels per variant, updated immediately on each query

Because the Storefront MCP is public and unauthenticated, any MCP-compatible AI agent can access it without requiring your customers to go through any additional authorisation step. A customer in Claude doesn't need to 'log in to your store', they simply ask a product question and the agent queries the publicly accessible MCP endpoint.

Customer Accounts MCP: Post-Purchase AI Access

The Customer Accounts MCP is an authenticated connection designed for interactions that require a customer to be identified, order tracking, return requests, subscription management, loyalty point balances, and account details.

Unlike the Storefront MCP, this connection requires the customer to authenticate before an AI agent can retrieve their personal data. The authentication flow uses OAuth 2.1, meaning the customer grants the AI agent permission to access their Shopify account data, similar to how you might grant a third-party app access to your Google account.

What the Customer Accounts MCP Handles

• Order history, list and retrieve past orders with status, items, and fulfilment details
• Order tracking, live tracking status and estimated delivery for active orders
• Return requests, initiate and manage return and exchange requests through the agent
• Subscription management, view, pause, or modify subscription products
• Account details, name, address book, and saved payment methods
• Loyalty and rewards, points balances and redemption options where applicable

The Customer Accounts MCP transforms post-purchase customer service for AI-capable stores. Instead of a customer navigating to your store, logging in, and finding their order history manually, they can simply ask Claude 'Where is my order from Any Car Seat Covers?' and the agent retrieves live fulfilment data on their behalf, after a one-time authorisation.

Can You Run Both Simultaneously?

Yes, and for a complete AI-ready Shopify store, you should. The two MCP connections are designed to complement each other. They cover different phases of the customer relationship and different categories of data, so there is no conflict or redundancy between them.

A customer's journey might start with the Storefront MCP, discovering and purchasing a product through an AI assistant, and then shift to the Customer Accounts MCP post-purchase for order tracking and, if needed, a return. Both connections are needed to serve that journey completely through AI.

Which One to Implement First?

For most Shopify stores, the Storefront MCP delivers more immediate commercial impact and should be implemented first. It directly affects AI-mediated product discovery, the fastest-growing new customer acquisition channel in 2026. It's also easier to test and validate because it requires no customer authentication.

Implementation Considerations for Each

Storefront MCP Setup Requirements

• Headless sales channel installed in Shopify admin
• Storefront API access token with product read, cart write, and page read permissions
• MCP server endpoint configured and accessible over HTTPS
• Product data quality audit, titles, descriptions, variants, and metafields must be structured for AI querying
• Discovery files updated (llms.txt, ai.txt) to expose MCP endpoint to crawling AI agents

Customer Accounts MCP Setup Requirements

• New Customer Accounts (Shopify's updated account system) must be enabled, legacy accounts don't support MCP
• OAuth 2.1 authorisation flow configured for the AI agent connection
• Customer Account API access scopes defined, orders, returns, subscriptions as applicable
• Customer data handling documented for POPIA / GDPR compliance
• Testing across supported AI clients before announcing to customers

The Bigger Picture: MCP as a Shopify Competitive Moat

Whether you implement Storefront MCP, Customer Accounts MCP, or both, the strategic value is the same: you're making your Shopify store a first-class citizen in the AI agent ecosystem at a point where most stores haven't started. That timing advantage compounds over months, every AI interaction that references your store builds retrieval confidence, and first-mover AI citations are genuinely difficult for later adopters to displace.

eComerfy MCP implements both Shopify MCP connections for South African and international Shopify stores, including endpoint configuration, product data quality optimisation, OAuth authorisation flow, and ongoing monitoring of AI-mediated interactions. Contact us to find out what MCP readiness looks like for your store.