Example 1: finding the right product
Customer: "I need a rain jacket for cycling, not a heavy winter coat." Agent: "For cycling, I would start with the Commuter Shell because it is lightweight and has a longer back. Do you need room for a layer underneath?" The reply reflects the use case and asks only one question that can change the recommendation.
Example 2: comparing two products
Customer: "What is the difference between the 20L and 30L pack?" Agent: "The 20L is designed for day trips and is lighter. The 30L adds capacity for extra clothing or overnight gear. Both have the same laptop sleeve." Comparison works when the shopper can see the deciding attribute without reading two full product pages.
Example 3: choosing a variant
Customer: "Which size should I order?" Agent: "The size guide lists a 92 cm chest for medium and 100 cm for large. If you are between sizes, the fit notes recommend large for a looser layer. Would you like the size guide link?" Variant advice must use the store's real sizing information.
Example 4: shipping before checkout
Customer: "Can this arrive before Friday?" Agent: "Standard delivery is usually 3 to 5 business days after dispatch, but I cannot guarantee a Friday arrival from the information available here. Your checkout will show the available delivery options for your address." Honest uncertainty is more useful than a confident promise that creates a support problem later.
Example 5: order status
Customer: "Where is my order?" Agent: "Your order has been dispatched and the tracking link shows the latest courier update here. If the tracking does not change within the stated delivery window, our team can investigate it." The answer should use available order data and name the next step when the status stops moving.
Example 6: returns and exchanges
Customer: "Can I exchange this for another size?" Agent: "Our policy allows exchanges for unused items within 30 days. Start the request from your order email, and the team will confirm whether the requested size is available." Link to the actual policy and avoid treating a general rule as an approval of this specific order.
Example 7: an unavailable product
Customer: "Do you have the black version?" Agent: "The black variant is not available in the catalog right now. The charcoal version has the same dimensions and is currently listed. I can show it if that helps." Do not hide the stock limitation or substitute a product with different properties.
Example 8: abandoned cart follow-up
Customer: "I left checkout because I was not sure about the shipping cost." Agent: "Shipping options and cost are calculated at checkout for the delivery address. I can answer questions about the product or connect you with the store team if the estimate looks incorrect." A recovery message should remove uncertainty, not pressure someone who has not given marketing consent.
What makes a live chat example useful
The best examples show the reasoning and the boundary: what information the agent used, what it could not verify, and when a person should take over. Dejor brings product discovery, customer support, cart assistance, and human handoff into a Shopify storefront conversation. Explore the Shopify AI chatbot page to see how these exchanges fit a real store workflow.
Explore this workflow in Dejor.
See how shopify ai chatbot and shopping assistant fits into your store, including setup steps and common questions.