E-commerce
July 1, 2026
After an in-store, telephone, or chat exchange, an advisor can prepare a cart for the customer. This is convenient, but it also raises questions: where is the link, what does the cart contain, why does the price differ, how long remains valid?
The chatbot can handle these simple requests if the cart is properly identified. It must explain, reassure, and guide towards completion without randomly altering the content.
This guide explains how to structure this journey to avoid lost carts and price misunderstandings.
Summary
Why does an advisor cart generate questions?
A shopping cart created by an advisor blends two worlds: human advice and online checkout. The client may trust the advisor, but find themselves alone in front of a link, a total, or a availability that has changed.
The most frequent requests concern the unreceived link, the contents of the cart, expiration, price differences, or the next step to pay.
The chatbot must extend the advice, not restart the entire sale.
Convert over 2,000 customers on average per month with Qstomy.
The world’s 1st Shopify AI dedicated to customer conversion
Empowering 200+ e-commerce merchants
What is the difference with a saved or shared cart?
A saved cart is usually created by the customer. A shared cart can come from a friend or a social link. A consultant cart is prepared by a person from the brand, often after a recommendation.
This origin changes the customer's expectation. They don't just want to find products: they want to understand if the cart matches what the advisor proposed to them.
Which cases should the bot recognize?
The bot must identify whether the customer is looking for the link, checking the content, contesting the price, requesting a modification, or wanting to know if the cart is still valid.
If the problem concerns a complex business recommendation, the bot can summarize the context and transfer to the advisor or support. If the question is logistical, it can often answer directly.
Which data to connect?
The chatbot must be able to retrieve the cart using an email, a token, a link, or a reference provided by the advisor. It must read the products, variants, quantities, prices, availability, and expiration date.
It must also know whether the cart can be modified or if it must remain compliant with a validated proposal. This distinction prevents breaking an offer prepared by an advisor.
Which rules apply?
The bot must not promise an old price if the cart has expired or if stock levels have changed. It must explain the situation transparently: "The cart was prepared on [date]. Some elements may change until the order is finalized."
It must also avoid replacing a product without validation when the recommendation depended on personalized advice.
If the customer disputes an oral promise, the bot must transfer to a human with the maximum amount of context.
Which flow to follow?
The flow must allow retrieving the cart and helping the customer finalize their purchase.
Identify the cart via email, link, or reference.
Confirm the main content: products, variants, and quantities.
Verify the price, discounts, and availability.
Explain the validity period or expiration.
Resend the finalization link if everything is consistent.
Escalate if the customer disputes the advice or the price.
Which messages should be used?
For a missing link: "I can help you find the cart prepared by the advisor. Could you confirm the email address used during the conversation?"
For the content: "Your cart contains [summary]. I recommend checking the size and quantity before payment."
For the expiration: "This cart was valid until [date]. If you wish to reactivate it, I can forward the request to our team."
When to transfer?
Transfer is necessary if the customer mentions a promise made by an advisor, a discount that is not visible, a product substitution, or a major modification request.
The bot must transmit the basket link, current content, displayed price, creation date, and the customer's exact request.
Which KPIs should be monitored?
Track the rate of finalized advisor carts, resent links, expired carts, price disputes, and transfers to advisors.
If many carts expire without converting, it may be necessary to improve the delay, the email reminder, or the clarity of the sent link.
Which mistakes should be avoided?
Avoid saying the price is guaranteed if that is not the case. Also avoid modifying a cart prepared by an advisor without explaining the impact on the recommendation.
The chatbot must remain factual: current content, current price, current validity, and next step.
How can Qstomy help?
Qstomy can retrieve a prepared cart, explain its contents, and guide the customer to the checkout with the context of the recommendation.
If the question relates to price, a promise, or a sensitive modification, the bot can transfer to a human with the cart details.
Explore the AI sales agent, the AI support, or request a demo.
ADVCRTBOT Checklist (8 steps)
Sync ADVCRT-MAP #983: RAG bot cart content skus index
Policy ADVCRTBOT-SUP: 6 rules REGISTRY-GATE CONTENT-GROUNDED NO-PRICE-INVENT
8 intents bot_advcrt_*: flow ACB-1 to ACB-8
5 templates TPL-ADVCRT-*: CONTENT LINK PRICE CHECKOUT HANDOFF
Advisor cart widget: cart_detect post-send-cart entry
draft_order_id verify: live total delta API test
Red team 10 carts: invent content price link out of map test
KPI Dashboard: advcrt_bot_* section 9 + delta advcrt_
FAQ
Difference #983?
#983 = agents verify resend price ops modify execute. #984 = bot tier 1 content widget.
Does the bot invent cart content?
No, out of ADVCRT-MAP. NO-CONTENT-INVENT. Handoff if unknown.
Difference with STOREADVBOT #982?
#982 = advice recommendation without cart. #984 = draft cart send-cart.
Difference with SAVEDbot?
SAVED = online customer cart. #984 = POS seller prepared cart.
Checkout price difference?
Bot quotes cart_total_ref + live delta. Persistent → handoff #983 PRICE.
Go further
This week: sync ADVCRT-MAP #983, advisor cart widget, CONTENT LINK CHECKOUT templates, measure advcrt_bot_deflect.
Enzo
July 1, 2026
