E-commerce
July 1, 2026
"The advisor prepared a cart for me, but I never received the link." "The price at checkout does not match what I was told in-store." "The cart expired before I could order." These incidents break the continuity between in-store advice and online purchasing.
The support for advisor-created carts must find the exact cart, verify its contents and price, and then provide the customer with a clear next step. This journey involves a draft cart or a link sent by a salesperson; it differs from simple in-store advice without a saved cart.
Guide #983 presents policy ADVCRT-SUP, flow AC-1 to AC-8, and matrix ADVCRT-MAP. It complements the future advisor cart bot (#984), which will be able to retrieve the link and explain the content before any human intervention.
Summary
Why can a basket prepared in-store block an online purchase?
An advisor can build a cart in Shopify POS or send a draft cart by email or SMS. Once back home, the customer may no longer find the link, discover that an item has been removed, notice a price discrepancy, or learn that the cart has expired. Without ADVCRT-MAP, the agent does not have the necessary information to accurately reconstruct the proposal.
The five most frequent challenges
Missing link: the email or SMS was not received or was filed elsewhere.
Unclear content: the customer does not understand the selected products or the advisor's recommendation.
Price discrepancy: the checkout total differs from the amount announced or saved when created.
Expired cart: the draft cart has exceeded its validity period.
Impossible modification: the customer wishes to add, remove, or replace an item before paying.
Omnichannel example
A beauty brand present in 38 stores received nine advcrt_ tickets per month. After the creation of ADVCRT-MAP, the resolution rate reached 86% and the conversion rate of sent carts into orders increased by 33% in five weeks.
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
How do you distinguish ADVCRT from other types of carts?
A cart can be created by an advisor, saved by the customer, shared with a relative, or simply lost between two devices. The source of the cart determines the available information and the procedure to follow.
Which path to use?
#983 ADVCRT: an advisor has prepared a cart and sent a finalization link.
STOREADV #981: the seller has given advice without creating a cart.
EXPIRED: a classic web cart has expired.
LOSTCART #751: the cart disappeared during a change of device.
Bot #984: the bot finds the cart and guides simple requests.
STOREADV #981 concerns a recommendation without a saved cart. ADVCRT #983 concerns a cart already created by the advisor and associated with a link or an identifier.
What guide #983 provides
ADVCRT combines the ADVCRT-SUP policy, the ADVCRT-GATE tree, the response macros, the advisor cart register, and the KPI advcrt_cart_to_order_rate.
Which advcrt_* typologies should be used?
The typology must indicate what actually prevents the customer from completing their purchase: link, content, price, expiration, stock, or modification.
The eight ADVCRT typologies
advcrt_link_missing: the link sent by email or SMS was not received.
advcrt_link_expired: the draft cart has exceeded its validity period.
advcrt_content_explain: the customer wants to understand the selected items.
advcrt_price_mismatch: the current total differs from the announced or saved amount.
advcrt_item_removed: an item was removed due to an out-of-stock situation or a stock update.
advcrt_modify_ask: the customer wants to modify the prepared cart.
advcrt_checkout_guide: they need help finalizing the order.
advcrt_ops_flag: the ticket reveals an issue in the send-cart program.
Use the tags advcrt, advisor_cart, send_cart, and draft_order. CONTENT-GROUNDED requires using cart_line_copy and cart_skus from the registry.
What ADVCRT-SUP rules govern the responses?
ADVCRT-SUP prevents the agent from rebuilding a memory cart or confirming a price that is not documented.
The six ADVCRT-SUP rules
CART-REGISTRY-FIRST: look up the cart in ADVCRT-MAP before responding.
CONTENT-GROUNDED: explain the content using cart_line_copy and cart_skus.
PRICE-VERIFY: check the draft_order_id in Shopify before explaining any discrepancy.
NO-PRICE-INVENT: do not promise any price that is absent from cart_total_ref or a validated policy.
LINK-RESEND-VERIFY: verify the sending and the address before concluding that the customer lost the link.
Handoff #984: pass the necessary fields to the bot to display the cart and the next step.
The minimal registry of an advisor cart
advcrt_id: identifies the cart in the program.
advisor_id_ref and store_location_ref: indicate the advisor and the store.
cart_url, cart_token_ref, and draft_order_id: allow retrieving the Shopify link.
cart_skus and cart_line_copy: describe the selected items.
cart_total_ref and price_at_creation: record the total at the time of creation.
cart_offer_code and cart_ttl_days: specify the offer and the validity duration.
advisor_note_copy and modify_policy_copy: provide recommendations and modification rules.
staff_brief_url: links to the procedure shared with the retail teams.
What path should be followed to resolve an ADVCRT ticket?
The AC-1 to AC-8 flow makes it possible to find the correct cart, verify the data live, and guide the customer without recreating an approximate proposal.
The eight AC steps
AC-1 — Collect: identify the request, the store, the email address, the link, or an order reference.
AC-2 — Retrieve the cart: consult CART-REGISTRY using the token or the draft_order_id.
AC-3 — Verify in Shopify: check SKUs, total, and stock in real time.
AC-4 — Classify: select link, content, price, expired, modify, or checkout.
AC-5 — Respond: use the ADVCRT macro powered by cart_line_copy.
AC-6 — Resend and guide: resend the cart_url and explain the checkout steps.
AC-7 — Modify or escalate: apply modify_policy_copy or create a retail ops_flag.
AC-8 — Document: record the resolution and ADVCRT KPIs.
An advcrt_link_missing request should ideally be resolved in a single interaction after verifying the address and resending the cart_url.
Which ADVCRT macros should be used?
Macros must extract data from the registry and indicate a clear action to the customer.
ADVCRT Library
ADVCRT-LINK-01: "Here is the cart prepared by our advisor from {{store_location_ref}}: {{cart_url}}. Its reference is {{advcrt_id}}."
ADVCRT-CONTENT-01: "The cart contains: {{cart_line_copy}}. The associated references are {{cart_skus}}."
ADVCRT-PRICE-01: "The total recorded during creation was {{cart_total_ref}}. The current total is {{total live Shopify}}. Here is the origin of the discrepancy: {{delta expliqué}}."
ADVCRT-EXPIRED-01: "This cart was valid for {{cart_ttl_days}} days. Here is how to obtain a new link: {{procédure}}."
ADVCRT-OOS-01: "The item {{sku retiré}} is no longer available. We can offer you {{alternative}}."
ADVCRT-MODIFY-01: "Here are the possible modifications for this cart: {{modify_policy_copy}}."
ADVCRT-CHECKOUT-01: "Open {{cart_url}}, apply the code {{cart_offer_code}} and then finalize the payment."
ADVCRT-DONE: "Here is the summary of ticket {{id}} and the applied solution: {{résolution}}."
How to use ADVCRT-GATE and coordinate retail?
ADVCRT-GATE prevents recreating a cart without verification or honoring an oral price that does not appear in any register.
ADVCRT-GATE Decision Tree
The customer mentions a link or a shop: search for the cart then verify it in Shopify.
The link is missing: check the address and resend cart_url with LINK-01.
The content is unclear: use CONTENT-01 with cart_line_copy and cart_skus.
The price is different: compare cart_total_ref to the live total and explain the delta with PRICE-01.
The cart has expired: use EXPIRED-01 and the authorized recreation procedure.
The customer wants to modify the items: apply MODIFY-01 and modify_policy_copy.
No cart exists: redirect to STOREADV #981.
The cart calculation is incorrect: use the cart calculation error guide.
Launch ritual for the send-cart program
Before the launch, validate cart_ttl_days, share modify_policy_copy with the shops, and test cart_offer_code in Shopify POS. Then, synchronize ADVCRT-MAP with bot #984.
Which KPIs and quality controls should be monitored?
The ADVCRT KPIs measure the send-cart program's ability to transform advice into an order without creating content or pricing errors.
The Five Key Indicators
advcrt_resolution_rate: share of tickets resolved without invented prices or content.
advcrt_cart_to_order_rate: number of orders originating from cart_url relative to carts sent.
advcrt_link_resend_success: share of links successfully resent after a link_missing report.
advcrt_price_delta_rate: share of price tickets in which the discrepancy has been verified and explained.
advcrt_price_invent_violation: number of promised prices out of range, with a target equal to zero.
Prepare the handoff to bot #984
Prioritize advcrt_content_explain, advcrt_checkout_guide, and advcrt_link_missing. The CONTENT-GROUNDED guardrail must enforce the use of cart_line_copy and cart_url from ADVCRT-MAP.
How do you identify baskets that belong to a different process?
Not all cart links come from an advisor. Identify their origin before using ADVCRT.
Cart saved by the customer
Use the saved cart guide when the cart was created directly by the customer online.
Shared cart
Use the shared cart guide when a relative has sent a sharing link.
Expired web cart
Use the expired cart guide when the cart is not part of the send-cart program.
Order already placed in store
If the purchase has already been completed in Shopify POS, use the omnichannel POS support guide.
How to train agents in twenty minutes?
The training must teach how to consult CART-REGISTRY, to distinguish ADVCRT from saved or shared carts, and to never invent a price.
Four exercises
Ticket A: the link is missing. The agent verifies the address and sends back cart_url with LINK-01.
Ticket B: the customer does not understand the products. The agent uses CONTENT-01.
Ticket C: the total has changed. The agent compares cart_total_ref with the live price and explains the discrepancy.
Ticket D: the cart has expired. The agent applies EXPIRED-01 and the recreation procedure.
How does Qstomy structure ADVCRT in your stack?
On Shopify, Qstomy can detect advcrt_* intentions, retrieve the cart from the registry, and use the LINK, CONTENT, PRICE, and CHECKOUT macros.
The Three Bricks
Routing: distinguishes between advisor_cart, saved_cart, and shared_cart.
Advisor cart registry: centralizes the link, SKUs, total, validity period, and modification rules.
Bot #984: displays content, explains the next step, and hands off complex cases.
In the example of the beauty brand with 38 stores, implementing the registry led to an 86% resolution rate and a 33% increase in the conversion rate of carts to orders.
Discover AI customer support or request a demo.
ADVCRT Deployment FAQ and Checklist
FAQ
Should we honor a price announced orally but missing from the register?
No. Check cart_total_ref and the live total, then explain the discrepancy. Any exception must follow a validated policy.
How does this differ from STOREADV #981?
STOREADV handles consultation without a created cart. ADVCRT handles a draft cart or a link already sent.
How does this differ from a saved cart?
A saved cart is created online by the customer. ADVCRT is prepared by an advisor.
How does this differ from a shared cart?
A shared cart is shared between individual customers. ADVCRT is sent by a salesperson or a store.
What is the role of bot #984?
It processes simple content, link, and checkout requests. Price discrepancies or complex modifications remain managed by agents.
Eight-Step ADVCRT Checklist
Validate ADVCRT-SUP and create the ADVCRT-MAP template.
Test cart_url and cart_token_ref for each shipment from Shopify POS.
Validate cart_line_copy and cart_skus when creating the cart.
Synchronize cart_total_ref and cart_ttl_days with the Shopify draft order.
Share modify_policy_copy with retail teams.
Test cart_offer_code at checkout on mobile.
Synchronize bot #984 with CONTENT-GROUNDED.
Train agents using the exercises in section 10.
Going Further
Enzo
July 1, 2026
