How Rep AI Personalizes Conversations Using Your Returning Customer's Last Order

Introduction

When a returning customer visits your store, Rep AI now sees more than just that they bought something — it sees what they bought. Product names, categories, brands, variants, prices, and refund status from their most recent order are all available to your AI in real time.

This unlocks three things you couldn't do before:

• Personalized proactive greetings — "How's the Hydrating Face Cream you picked up last month?"

• Purchase history Q&A — your customer can ask "what did I buy last time?" and the AI answers directly

• Smarter AI instructions — write rules like "If the customer bought from our skincare line, suggest the new serum" and the AI will actually act on them

There's nothing to install, no setting to toggle. The data flows in automatically from Shopify, and your AI is already using it.

What Your AI Now Knows About Each Past Purchase

For every product in a returning customer's most recent order, Rep AI has access to:

Fields that aren't set in your Shopify store are simply skipped — the AI works with whatever is available.

How Rep AI Uses This Information

Personalized Proactive Greetings

When a returning customer lands on your store, the AI can reference what they bought instead of opening with a generic line.

• Before: "Welcome back! How can I help today?"

• Now: "Welcome back! How's the Hydrating Face Cream working out for you? We've got a new serum that pairs perfectly with it."

The AI decides whether and how to reference past purchases based on the conversation context — it won't sound robotic or sales-y.

Answering "What Did I Buy Last Time?"

Customers often forget what they ordered, especially for repeat replenishment or gift purchases. Your AI can now answer these questions directly without escalating to support:

• "What did I order last time?"

• "What size did I get?"

• "Did I buy the lavender or the unscented one?"

The AI only references items it has actual data for — it won't make up product details.

Smarter AI Instructions (Your Biggest Lever)

This is where you take the wheel. You can now write instructions in your AI Training settings that reference any product attribute from your customer's purchase history.

Examples you can write today:

• "If the customer previously bought anything tagged 'skincare', proactively suggest the new Vitamin C Serum."

• "If the customer bought a gift card last time, ask if they'd like help finding the right product to use it on."

• "If the customer bought from the 'GlowLab' brand, mention our upcoming GlowLab restock."

• "If the customer's last order was over $200, offer them early access to our VIP collection."

• "If the customer bought from the 'vegan' line, only recommend products tagged 'vegan' going forward."

You can combine multiple attributes too — "If the customer bought a 50ml size of any skincare product, suggest our 100ml subscription option."

Matching is case-insensitive. "Skincare", "skincare", and "SKINCARE" all match the same instruction. And the AI checks every item in the order, not just the first one — so a 10-item order with one matching product will still trigger your instruction.

Refund Awareness

If a customer returned an item, the AI knows. It won't proactively reference returned items as a positive purchase — no awkward "How's the dress you returned?" moments.

Partial refunds work too. If a customer bought 3 units of something and returned 1, the AI sees both numbers and behaves accordingly.

What This Doesn't Do

• Brand new visitors don't have a last order, so this only applies to returning customers who have made at least one purchase.

• Only the most recent order is loaded — older order history isn't included. (This keeps the AI fast and focused.)

• Very large orders are summarized, not loaded in full. If a customer bought a long list of items, the AI sees the most relevant ones rather than every single line item.

• Data comes from your Shopify store — if a field isn't set on the product in Shopify (e.g., no product tags), the AI simply won't have it available for instructions.

Tips for Getting the Most Out of This

1. Audit your product tags and product types in Shopify. The richer your taxonomy, the more powerful your AI instructions can be. If most of your products lack tags or have generic types, the AI has less to work with.

2. Rewrite generic instructions to use product context. A rule like "Always upsell to returning customers" is less effective than "If the customer bought from our skincare line, suggest the moisturizer companion."

3. Use the brand/vendor field for cross-brand stores. If you carry multiple brands, you can write brand-specific instructions that wouldn't have worked before.

4. Test the experience. Place a test order, return as a logged-in shopper, and see how the AI greets you. The personalization is most visible in the first message.

5. Combine with merchant instructions you already use. Product-aware rules layer on top of your existing AI Training — they don't replace it.

Troubleshooting

• The AI isn't referencing my customer's last order in greetings.
  Confirm the customer is logged in or otherwise identifiable to your store. Anonymous browsers don't have order history attached.

• My instruction isn't triggering.
  Check that the product attribute you're referencing is actually set in Shopify (look at the product in your Shopify admin). If productType is blank for that product, an instruction matching on product type won't fire for it.

• The AI mentioned an item the customer returned.
  Refund data flows in from Shopify. If your store's refund records are incomplete in Shopify, the AI may not know about it. Make sure refunds are processed through Shopify's refund flow, not handled out of band.

• A customer asked "what did I buy?" and the AI gave incomplete information.
  Very large orders are truncated to keep the AI responsive. For the full order list, customers can check their account page or order confirmation email.