How to automate “Where is my order?” emails on Shopify

“Where is my order?” (support teams call it WISMO) is usually the most common question a Shopify store gets, and the most repetitive. The answer almost always exists already, in Shopify and in the carrier’s tracking. Someone just has to go and fetch it, again and again.

That makes it the best first thing to automate. Here’s how we’d approach it, step by step.

1. Measure it first

Before building anything, tag a week or two of tickets. You want to know:

• How many tickets are order-status questions?
• Which kinds? “Not shipped yet”, “stuck in transit”, “says delivered but it isn’t here” and “I need to change my address” need very different answers.
• How long does each one take your team today?

This gives you a baseline, and it tells you which cases are safe to automate and which always need a person.

2. Fix the cheap stuff

Some WISMO tickets are a symptom of missing information. Before you add AI, check that:

• Shipping confirmation emails include a working tracking link.
• Your order status page shows tracking clearly, and customers can find it.
• Your shipping policy states realistic dispatch and delivery times.

These fixes cost almost nothing and can noticeably shrink the problem on their own.

3. Connect the data

To answer properly, the automation needs the same facts your team looks up:

• The order. The Shopify Admin GraphQL API lets you find an order by number or customer email and read its fulfillment status, line items and shipping address.
• The tracking. Each fulfillment carries tracking details (carrier, tracking number and URL). For the live status and estimated delivery date, connect a tracking service or the carriers’ own APIs.
• The conversation. Your helpdesk (Gorgias, Zendesk, Help Scout or plain Gmail) provides the incoming question and somewhere to put the reply.

Keep this lookup in ordinary code, not in the AI. The AI should never guess an order number or a delivery date. It should be handed the facts.

4. Write down the rules

Decide what should happen in each situation before any AI is involved. For example:

Situation	What happens
Not shipped, within your dispatch window	Reassure, give the expected dispatch date
In transit and on time	Share tracking and the estimated delivery date
Delayed past a set number of days	Apologize, share tracking, flag to the team
Marked delivered but not received	Always goes to a person
Address change or cancellation	Always goes to a person
Order not found	Politely ask for the order number

This table is your policy. It belongs to you, not to the model.

5. Let the AI write, not decide

With the facts gathered and the situation identified, the AI’s job is narrow: write a friendly, specific reply in your brand voice. Give it the order facts as structured data, a few example replies your team is proud of, and firm instructions:

• Only use the facts provided.
• Never promise refunds, discounts or reshipments.
• If anything is unclear, say a teammate will follow up.

Narrow jobs like this are where AI is most reliable.

6. Roll it out with a person in the loop

For the first couple of weeks, post the AI’s replies as internal notes or drafts in your helpdesk. Your team reads each one, edits if needed, and sends. Track how often they change anything.

When the edit rate for a category is close to zero, say “in transit and on time”, let that category send automatically. Keep the sensitive ones (delivered but not received, anything angry) with a person for good.

7. Keep measuring

Watch a handful of numbers every week:

• First response time for order-status questions
• The share resolved without a person
• How often customers reply again (a sign the answer didn’t help)
• Customer satisfaction on those tickets

Build it or buy it?

Several helpdesks now include AI agents that can answer simple order questions, and if one fits your store, use it. A custom build earns its keep when your data lives in more places (a 3PL portal, an ERP, several stores), when your policies are unusual, or when you want full control over what the AI can and can’t do.

If you’d like a second opinion on where WISMO sits in your support queue, our free Shopify AI audit ranks it against your other automation options by hours saved.