Shopify Support Automation Without Losing the Human Touch: 7 Tickets to Automate and 5 to Keep for Your Team

Keeping the human touch does not mean answering every ticket yourself. It means a person handles the few tickets where a person makes the difference. This guide shows you how to decide, one ticket type at a time.

The 4-question test for any ticket

Use this test before you read the lists. It works on any ticket type, including ones not covered below.

• Is the answer in Shopify or in a written policy? If the order data or your policy page has the answer, an AI agent can find it.
• Is the action small and reversible? Changing an address before fulfillment is easy to undo. A large refund is not. The AI only acts within the action limits you allow in its permissions. Anything above them goes to your team.
• Is the customer usually upset on this type of ticket? Mild frustration is fine. Anger, fear or distress need a person.
• Does it need judgment beyond your rules? If your policy does not cover the case, someone has to make a business decision.

Automate the ticket type if you answer "yes" to questions 1 and 2 and "no" to questions 3 and 4. Any other combination goes to your team.

7 tickets to automate

• Where is my order: the AI needs order and tracking data. It replies with the status and the tracking link.
• Address change: the AI needs the order status and your cutoff rule. It updates the address if the order is not yet fulfilled.
• Order cancellation: the AI needs the order status and your cancellation rule. It cancels if the order is not yet fulfilled.
• Return or exchange within policy: the AI needs the order date, your return window and your product rules. It confirms eligibility and next steps.
• Product questions: the AI needs your product pages and knowledge base. It answers on size, material, care and compatibility.
• Shipping and policy questions: the AI needs your shipping and return policies. It answers on delivery times, countries and conditions.
• Discount code issues: the AI needs your promo rules and your permissions. It explains the issue, or issues a code if you allow it.

1. "Where is my order?" The answer is in the order: status, carrier and tracking link, so an agent connected to Shopify can reply with the exact status. The full setup is in our guide on how to automate WISMO tickets on Shopify, linked at the end of this article.

2. Address changes before fulfillment. The rule is simple. If the order has not shipped, the address can change. If it has shipped, it cannot. The AI checks the status, then updates the address or explains why it is too late.

3. Cancellations before fulfillment. This works the same way. You set your rule once, for example "cancel if not fulfilled." The agent then applies it to every request.

4. Returns and exchanges within policy. If the item is inside the return window and in a returnable category, the answer is yes. The AI confirms the customer can return it and sends the steps. Requests outside the window belong in the "keep" list below.

5. Product questions. "Does this run small?" "Is it machine washable?" "Will this fit my model?" The answers are in your product pages and size guides. If they are written down, the agent can use them. If they are not, write them once in your knowledge base and the gap is closed.

6. Shipping and policy questions. Delivery times, countries you ship to, return conditions, how exchanges work. These are policy answers, and getting one wrong costs little.

7. Discount code issues. A code that will not apply, an expired code, or a customer who just missed a sale. You decide whether the AI can issue codes. If you allow it, it can solve the issue directly. If you do not, it explains the problem and passes the request to your team.

5 tickets to keep for your team

On these tickets, a person protects the customer relationship. Keep them out of the rules and permissions you give the AI. Anything outside those rules goes to your team with the context attached.

1. Angry or distressed customers. A gift that arrived broken the day before a birthday. A customer who says they will post a review. The first reply should admit the problem in a human voice, before offering any fix. Do not write a rule that lets the AI settle these complaints.

2. Lost, damaged or wrong items. These often need photos, a carrier claim or a stock check. The fix depends on the details: reship, refund or partial refund. Each case is different, so a person should own it.

3. Chargebacks, disputes and suspected fraud. A bank dispute or a suspicious order carries financial and legal weight. The wording of a reply can matter later. Keep these with your team.

4. Exceptions to your policy. A return on day 45 when your window is 30 days. A refund on an item you do not normally refund. A loyal customer asking for a favor. These are business decisions. Your rules and permissions define what the AI can do. Everything else goes to a person.

5. Safety, health and legal concerns. An allergic reaction, a product that broke and hurt someone, a legal threat. These tickets need a careful, documented human answer every time.

The gray zone: tickets that change category mid-conversation

Some tickets start simple and turn complex. A tracking question is routine until the customer asks for a refund because the parcel has not moved. A return request is routine until the customer describes a problem with the product.

That is why narrow rules matter more than the split. The AI handles only what your rules and permissions cover. When a request falls outside them, the ticket goes to your team with the full context. The ticket reaches your team with the order, the history and what has already been said.

Start with tight rules. Widen them once you have read enough handovers to trust the setup.

How to set this up on your store

• Export one month of tickets and tag each one with a type.
• Run the 4-question test on each type and mark it "automate" or "keep."
• Write the rules for the types you automate: cutoff times, return window, which products can be returned.
• Set the permissions. Choose the actions the AI may take: refunds, discount codes, cancellations, address changes. If you leave an action off, those requests go to your team with the context attached.
• Review the handovers weekly. If a ticket type keeps getting passed to your team, either your rules are missing something or it belongs in the "keep" list.

Not sure you are ready? Read when to automate customer support in a Shopify store. For other ways to lighten the load, see how to reduce Shopify support tickets without hiring. Both guides are linked at the end of this article.

How Relay handles the split

Relay works like a support employee who follows your store's rules. It reads the request, finds the order in Shopify and applies your policies. It then acts within the permissions you give it: refunds, discount codes, cancellations or address changes. Tickets outside those rules go to your team in the same ticket center, with the context attached.

You decide what it can do through the rules and permissions you set.

FAQ

Which Shopify support tickets are easiest to automate? Tickets where the answer is in your order data or a written policy. That covers order status, address changes and cancellations before fulfillment, returns within the window, product details, shipping questions and discount code issues.

Which customer service tickets should stay human? Angry or distressed customers, lost or damaged items, chargebacks and fraud, exceptions to your policy, and anything involving safety, health or legal risk.

What happens to tickets Relay is not allowed to handle? They go to your team in the same ticket center, with the order and the conversation attached. You decide what Relay handles through the rules and permissions you set.

Can an AI agent issue refunds on its own? Only if you allow it. If refunds are not in the permissions you give, every refund request goes to your team with the order and the conversation attached.