How Shopping Assistant recommends productsUpdated 3 days ago
Who can use this feature?
When a shopper starts a conversation, Shopping Assistant doesn't recommend products at random. It builds a picture of who the shopper is, what they're looking for, and how close they are to making a purchase — then uses that context to decide when and what to recommend.
In this article, you'll learn:
- How Shopping Assistant builds a picture of each shopper
- How Shopping Assistant decides what to recommend
- How recommendations are shared in conversations
- Managing which products Shopping Assistant recommends
How Shopping Assistant builds a picture of each shopper
Shopping Assistant is context-aware. It observes each shopper's activity during their visit to build an understanding of their interests and buying intent: how likely they are to make a purchase.
Browsing activity
Shopping Assistant tracks the following signals from a shopper's current session:
- Products viewed — which product pages the shopper has visited
- Page views — what pages they've explored, such as FAQs, reviews, or category pages
- Current page — which page they're currently browsing
- Cart contents — what items are currently in their cart
- Purchase history — what they've bought from your store before
Buying intent
Using these signals, Shopping Assistant assesses the shopper's buying intent — their likelihood of making a purchase. Buying intent can change during a conversation as the shopper asks questions, views products, or adds items to their cart.
| Stage | What it means | Key signals |
|---|---|---|
| Discovery | The shopper is browsing with no clear purchase signal | Vague questions, broad exploration, logistics questions with an empty cart |
| Interested | The shopper has clear interest in a product but hasn't committed | Mentions a specific product or category, asks about variants, sizing, or restocks |
| Ready to buy | The shopper has strong purchase intent | Cart contains products, asks about checkout or shipping, expresses intent to buy |
How Shopping Assistant decides what to recommend
Shopping Assistant combines what it knows about your products with what it observes about each shopper to decide when and what to recommend.
What Shopping Assistant knows about your products
Shopping Assistant draws on 3 sources of product information:
- Your Shopify catalog — product names, descriptions, and main product image, synced automatically from Shopify
- Your store website — product page content synced from your store's website, which may include details not in the Shopify catalog
- Custom product information — additional details you add manually in Gorgias at AI Agent > Products, such as materials, intended audience, or warranty details
To see what information Shopping Assistant has about a specific product, or to add custom information, see Manage your product information for AI Agent.
Shopping Assistant can also use guidance to inform its recommendations — for example, to flag that a product isn't suitable for a specific use case, or to prioritize a product for certain types of shoppers. Unlike the product information sources above, guidance provides instructions about how to use product knowledge, rather than facts about the products themselves.
Types of recommendations
Shopping Assistant can make 3 types of recommendation:
- Similar products — products that match something the shopper mentioned or is looking for: "Do you have something like the Nimbus 25?"
- Complementary items — products that go well with what the shopper is currently viewing: "Our extra durable laces pair well with this model and come in different colors"
- Popular or trending products — used when the shopper is casually browsing or when no products closely match their stated interest: "You might like the Nimbus 25, our best-selling shoe of the year"
How selling style affects recommendations
Your selling style affects how proactively Shopping Assistant makes recommendations and how it approaches different stages of a conversation. For example, an educational selling style means Shopping Assistant will ask more questions before recommending anything, while a promotional style means it will make recommendations earlier and more assertively.
To learn more about selling style options and how to configure them, see Customize your selling and discount strategy.
How recommendations are shared in conversations
Shopping Assistant shares recommendations differently depending on the channel where the conversation is happening.
- Chat — Shopping Assistant shares a visual Shopify product card. Shoppers can select the card to go to the product page or add the item directly to their cart.
- Email and SMS — Shopping Assistant shares a text link to the recommended product's page. It doesn't include a product image or card.
Out-of-stock products
If a product isn't available in a shopper's region, Shopping Assistant won't recommend it in chat conversations.
Shopping Assistant may still recommend or discuss an out-of-stock product if:
- The shopper's location isn't available
- The conversation is happening over email or SMS
- The shopper asks specifically about the product
- No in-stock products closely match the shopper's intent (in this case, Shopping Assistant will say the product is currently unavailable and won't include a product card)
To make sure Shopping Assistant is aware of inventory across your locations, update Gorgias's app permissions for your Shopify store.
Managing which products Shopping Assistant recommends
Shopping Assistant can recommend any product from your Shopify catalog. You can customize this by promoting specific products so Shopping Assistant prioritizes them, or by excluding products you don't want recommended.
To manage your promoted and excluded products, see Control which products Shopping Assistant recommends.
FAQ
Can Shopping Assistant recommend specific product variants, like a specific color or size?
Shopping Assistant recommends products at the product level, not the variant level. It uses a product's main image to understand what it looks like, so if a shopper asks for a product in a specific color and the main product image shows a different color, then Shopping Assistant may not recognize that the product is actually a match.
If your store carries products with visually distinct variants (such as different colors or styles), you can reduce the chance of mismatches by setting the most commonly searched or sold variant as the default product image in Shopify.
