AI Shoppers & Your Loyalty Program: Can They See Your Customer's Earned Rewards?

Hey everyone,

I’ve been following some really insightful discussions in the Shopify community lately, and one thread particularly caught my eye. It’s all about the fascinating, and sometimes frustrating, world where AI shopping assistants meet your store’s loyalty program. We’ve talked before about whether AI can even tell if your store has a loyalty program – that’s what we call “store-level visibility.” But what about the next level up? Can an AI shopper actually see what your specific customer has already earned? Like, that $15 reward sitting unused in their account?

That’s the exact question YangW_Stabile posed in a recent discussion, building on an earlier post. Imagine a scenario: a loyal customer has a $15 reward at Store B, which lists an item for $85. Store A sells it for $80. An AI assistant, without knowing about that $15 reward, would tell the customer Store A is cheaper. But the real deal? Store B is actually $70! This “identity-level visibility” gap is a huge deal for customer experience and conversion.

The Identity Gap: Why AI Struggles to See Personal Rewards

So, why is this so tricky? As ecom-4all pointed out in the thread, there's a fundamental “identity gap” in how Shopify's agent-facing surface works. When an AI assistant interacts with a store, it's largely anonymous. It sees your catalog, cart, checkout, discounts, fulfillment, and orders. But here’s the kicker: there's no customer lookup built directly into that surface. Identity usually gets resolved much later in the process, typically at checkout through Shop Pay. By then, any price comparison an AI might be doing has already happened.

The only real lever that crosses this anonymous barrier early on, ecom-4all noted, is a discount code. If you publish a code that the cart accepts without a login, that code can land in the AI’s price math. But those precious loyalty points, tied to an individual customer account, remain invisible.

Diving Deeper: Shopify's Own Customer Account API

Lumine, another expert in the discussion, chimed in with some crucial technical details about Shopify's first-party solution. Shopify does have a layer for this: the Customer Accounts MCP API. You can find it through a discovery call to /.well-known/customer-account-api, and it lives at {shop}/customer/api/mcp. But here’s the catch, and it’s a big one: it responds with a 401 (Unauthorized) until you present an OAuth 2.0 access token, obtained via an authorization code flow with PKCE, and with the scope customer-account-mcp-api:full.

This 401 isn't an error in the traditional sense; it's what kicks off the authentication flow. The problem is, this “gate” is per shop. For an AI assistant to proactively check for rewards at Store B while comparing it to Store A, it would need to already hold an authenticated token for Store B. Getting that token means a redirect and a consent screen – a pretty big hurdle to jump when the shopper hasn't even committed to Store B yet, or perhaps hasn't even mentioned it!

Building Solutions: What the Community is Doing

YangW_Stabile and the team at Stable Rewards have actually built their own MCP server to tackle this. They shipped it as a ChatGPT plugin and a Claude connector, and it does operate at the identity level. Once a shopper is recognized in the conversation, it can read a merchant’s program terms, see what a cart would earn, check a specific shopper’s balance, and even apply an existing reward to an order. The redemption part, YangW_Stabile confirmed, is fully proven and works end-to-end.

However, the “proactive” part – the AI assistant spontaneously surfacing “Hey, you have a reward here!” without being asked – is still a work in progress. Identity-level visibility exists once you ask for it; it doesn’t yet volunteer itself.

Critical Distinction: “Can't Verify” vs. “Zero Balance”

This is perhaps one of the most important takeaways from the entire discussion. Lumine highlighted that a 401 response from a shop where a customer has fifteen dollars banked looks exactly the same as a 401 from a shop with no loyalty program at all. If an AI system interprets the absence of a token as an absence of a balance, it will wrongly report “zero.”

Clickfromai, responding to Lumine, confirmed they handle this by keeping separate identity states:

• Identity not confirmed: Return “can’t verify.”
• Shop discovered but not connected: Offer the OAuth flow.
• Connected with confirmed zero balance: Return zero.
• Connected with a balance: Include it in the price comparison and redemption path.

This is a crucial best practice for anyone building these kinds of integrations.

Actionable Advice for Developers and Store Owners

If you're an app developer or an integrator working on AI shopping assistants for Shopify, here's what we've learned from the community:

1. Start with Discovery: Always perform the /.well-known/customer-account-api discovery check first to see if a shop supports the Customer Accounts API.
2. Cache Shop Capability: Once discovered, cache that shop's capability so you don't need to re-check constantly.
3. Manage Identity States Carefully: Implement distinct states for identity confirmation and balance status. Never assume a missing token or a 401 response means a zero balance; it means you “can’t verify.”
4. Be Strategic with OAuth: Avoid triggering the PKCE (Proof Key for Code Exchange) flow across every potential store. Only ask for customer consent (which initiates the OAuth flow) when the shopper explicitly selects that store, asks about rewards, or has connected it before. This respects user experience and minimizes friction.

For store owners, while the AI landscape for proactive loyalty visibility is still evolving, you can still make your loyalty program more visible to AI agents today. If you want member value weighed early in the shopping journey, consider publishing member-specific prices directly on product pages or offering discount codes that the cart will accept without a login. This provides a direct, unauthenticated signal that AI can pick up.

It's clear that bridging the gap between AI shopping assistants and personalized customer loyalty is a complex but exciting frontier. The community is actively pushing the boundaries, finding structural reasons for challenges, and building sophisticated solutions. As these AI tools become more prevalent, ensuring they can truly understand and act on a customer's earned rewards will be vital for delivering seamless, intelligent shopping experiences.