Marketing and ecommerce
WhatsApp abandoned-cart flow
Set up a WhatsApp abandoned-cart flow in Kanesh: consent, two templates, timing, replies in the inbox and how to read the results.
8 min read
Playbook
Answer order status without a person
Let the AI answer "where is my order?" with live Shopify data after verifying the customer, and hand off the cases it shouldn't handle.
For: Customer service7 min read
Goal
Customers who ask where their order is get an answer from live Shopify data in the same conversation, and the cases that need a person reach your team with the context.
What you'll have at the end
The AI answers order-status questions after verifying the customer; order changes, lost parcels and complaints go to your team with the reason.
On this page
Before you start
- Shopify connected to Kanesh, so orders, fulfilments, tracking and returns are readable.
- The channels where customers ask connected to the shared inbox: WhatsApp, Instagram, Messenger, email or web chat.
- Your shipping policy, delivery times by country and carrier information written down.
- Someone on your team who owns the conversations the AI hands off.
Set it up, step by step
Step 1
Give the AI your delivery facts
The AI reads the order in Shopify, but the rules around it (delivery times per country, what "in transit" means for your carrier, when a parcel counts as late) come from you. Put them in the knowledge base in plain sentences.
In Kanesh
- Knowledge base: upload the shipping policy, delivery times and a short FAQ on tracking.
Step 2
Verify before sharing order details
Kanesh only shares order details once the customer is verified. Tell customers in your FAQ and your first reply what they should have to hand, so verification is quick.
Step 3
Write the handoff rules for order questions
Hand off when the order can't be found, the tracking shows a problem or a lost parcel, the customer wants to cancel, change the address or get a refund, the customer complains, or asks for a person. Those actions are done by your team, not by the AI.
In Kanesh
- Handoff rules: list these situations. The AI passes the conversation with the reason and context.
Step 4
Route handoffs to the right people
Send order handoffs to the team that can act in Shopify, and set business hours and an SLA target so late replies raise an alert instead of going unnoticed.
In Kanesh
- Assignment to a team, business hours and SLA targets with alerts.
- Saved replies and macros for the answers your team repeats.
Step 5
Cut the question before it's asked
Proactive updates prevent a share of "where is my order?" messages. Kanesh can send the confirmation when an order is paid. For shipped, delayed or delivered updates, send the template from your own system or manually.
In Kanesh
- Flows: order paid trigger → send the order-confirmed template.
- Shipped and delay updates: inbound webhook trigger from your system, or a manual send.
Step 6
Review a sample every week
Read 20 order-status conversations a week: were the facts right, was the tone right, did the handoffs make sense? Fix the knowledge base or the rules, not individual answers.
What to measure
| Metric | How to read it |
|---|---|
| Resolved without a handoff or a reopen | Order-status conversations the AI closed and the customer didn't reopen. More honest than "conversations the AI touched". |
| Handoffs by reason | Lost parcels, order changes, no data found. Each reason tells you what to fix or who to staff. |
| First response time on handoffs | Against your SLA target. A fast AI answer followed by a slow human one is still a slow experience. |
| Reviewed accuracy | The share of your weekly sample with correct facts. Keep the sample and the result in a simple sheet. |
Common mistakes
- Sharing order details before the customer is verified.
- Letting the AI promise delivery dates the carrier hasn't given. Stick to what the tracking says.
- Leaving the knowledge base empty: without your policies, the AI has to hand off almost everything.
- Letting "cancel my order" wait in a general queue. Order changes are time-sensitive; route them first.
- Measuring only volume. A high share of AI answers means little if customers write back the next day.
What Kanesh doesn't do here
- The AI reads orders; it doesn't cancel, refund or edit them. Your team does that in Shopify.
- Kanesh has no built-in "shipped" or "delivered" trigger. Those updates come from your own system or a manual send.
Checked against Kanesh's features on September 30, 2026. Meta's rules change often; check the linked sources before you launch.
Templates for this playbook
Adapt the text to your brand. Meta reviews every template before you can send it.
- Utility
Order confirmed
Thank you, {{1}}. Your order {{2}} is confirmed. We'll message you here when it ships. If you have a question, just reply.
- Utility
Order shipped
Hi {{1}}, your order {{2}} has shipped with {{3}}. Your tracking number is {{4}}, and you can follow the delivery with the link below.
- Utility
Delivery delay
Hi {{1}}, your order {{2}} is taking longer than expected. The new estimated delivery date is {{3}}. We're sorry for the wait. Reply here if you have any questions.
- Utility
Delivered: all good?
Hi {{1}}, your order {{2}} shows as delivered. If anything is missing or isn't right, reply to this message and we'll sort it out.
- Utility
We have your answer
Hi {{1}}, we have an answer to your question about {{2}}. Tap below to continue the conversation.
More playbooks
- Marketing
Launch a WhatsApp campaign to opted-in customers
From audience to approved template to send speed: a WhatsApp campaign only to people who opted in, with replies answered and orders attributed.
9 min read
Free diagnosis
See what your conversations are leaving on the table.
We review up to 180 days of your WhatsApp history and send you a written diagnosis within 48 hours. Free, and nothing is connected until we've agreed in writing how your data is handled.