Configuring Upsell & Cross-Sell Recommendations
The Upsell Recommendations skill lets your AI Agent suggest related products during conversations. To get the best results, you can create upsell bundles that tell the agent exactly what to recommend and when.
Go to AI Agent → Skills, find Upsell Recommendations, and click Configure to manage your bundles.
What Are Upsell Bundles?
An upsell bundle is a rule that says: "When a customer is looking at these products, recommend those products." Bundles give you control over what the agent suggests instead of relying on general recommendations alone.
Bundle Types
Type | Best For | Example |
Frequently Bought Together | Products customers commonly buy in the same order | Running shoes + performance socks |
Complementary | Products that pair well but aren't always purchased together | Phone case + screen protector |
Upgrade Path | Higher-tier alternatives to what the customer is looking at | Basic plan → Pro plan, or Standard hoodie → Premium hoodie |
Creating a Bundle
Click Create Bundle in the configuration modal
Fill in the fields:
Name: A label for your reference (e.g., "Running Shoe Accessories")
Bundle Type: Choose from Frequently Bought Together, Complementary, or Upgrade Path
Trigger Products: The products that activate this bundle. When a customer is looking at or has these in their cart, the recommendations kick in. Select from your Shopify catalog.
Trigger Collections: Alternatively (or additionally), trigger on entire collections. If a customer is browsing anything in your "Running Shoes" collection, the bundle activates.
Recommended Products: The products the agent should suggest. You can add an optional reason for each recommendation (e.g., "Pairs well for extra arch support") to help the agent explain the suggestion naturally.
Priority: When multiple bundles match, higher priority bundles are recommended first. Use this when a customer's cart matches several bundles and you want to control which suggestions take precedence.
Recommendation Prompt (optional): Custom guidance for the agent on how to present these recommendations. For example, "Mention that buying both saves 15% with our bundle discount" or "Position as a gift add-on."
Click Save
Examples
Frequently Bought Together: Skincare Routine
Trigger Products: Hydra Boost Moisturizer
Recommended Products: Hyaluronic Acid Serum ("Customers love layering this under the moisturizer for extra hydration"), Gentle Foaming Cleanser ("Completes the routine")
Recommendation Prompt: "Suggest as a full routine, not individual add-ons"
Complementary: Phone Accessories
Trigger Collections: iPhone Cases
Recommended Products: Tempered Glass Screen Protector, MagSafe Charging Pad
Priority: 1 (high)
Upgrade Path: Subscription Tier
Trigger Products: Classic Coffee Subscription (12oz)
Recommended Products: Premium Coffee Subscription (12oz, single origin) ("Same price per cup, better beans"), Family Size Subscription (24oz) ("Best value for households")
Recommendation Prompt: "Only suggest the upgrade if the customer seems interested in quality or value"
How the Agent Uses Bundles
The agent doesn't force recommendations into every conversation. It looks for natural moments to suggest products based on what the customer is asking about or what's in their cart. When a bundle matches, the agent uses your recommended products and prompt guidance to make the suggestion feel organic.
If no bundle matches the conversation context, the agent can still make general recommendations based on your catalog. Bundles give you more control, but the agent is helpful even without them.
Tips
Start with your top 3-5 product pairings. Look at your order data for products that are frequently bought together and create bundles for those first.
Use recommendation prompts to guide the tone. "Suggest as a must-have accessory" produces a different message than "Mention casually if the customer seems interested."
Set priorities when bundles overlap. If a customer browsing running shoes could match both a "Running Accessories" bundle and a "New Arrivals" bundle, priority determines which products get suggested first.
Keep it natural. The agent works best when it has a few high-quality bundles rather than dozens of low-priority ones. Customers can tell when recommendations feel forced.
Test in the Playground. Create a test conversation about one of your trigger products and see how the agent presents the recommendations.
Archive seasonal bundles. If you created a bundle for a holiday gift guide, archive it after the season ends so the agent doesn't suggest holiday pairings in February.
💡Tip:
Still have questions?
Please feel free to reach out to our wonderful Support team at support@txtcart.ai or via Live Chat.