The short answer
Automate
- Routing: tag tickets by topic and send them to the right person or queue.
- Order-status answers: look up the order and the tracking, and reply with facts.
- Return eligibility checks: order date, item, condition rules, and your policy window.
- Refund prep: gather the ticket, order, and return receipt into one ready-to-approve case.
- Data entry between the helpdesk and the store: notes, tags, and status kept in sync.
Keep human
- Exceptions: anything the rules don't cover cleanly.
- Upset or escalating customers, and anything that needs judgment or tone.
- Refunds or credits outside your written policy.
- Anything that changes money without someone approving it first.
The rule of thumb: automate the gathering and the checking, keep the deciding. A system can collect every fact a refund decision needs in seconds. Whether a long-time customer gets an exception is still your call.
When a returns app or your helpdesk is enough
Start with what you already own. Shopify, returns apps, and helpdesks such as Gorgias and Zendesk all include automation, and for many brands it covers the job. If the work stays inside one tool, you probably do not need a custom system, and you do not need Cadence Ridge. Typical examples:
Customers request returns themselves
Shopify's self-serve returns lets customers request a return within the rules you set, and you approve or decline each one. Returns apps such as Loop are another option.
Tickets tagged and assigned
Gorgias rules can tag, assign, and auto-reply to incoming tickets. Zendesk triggers update tickets and send notifications when conditions you define are met.
Replies to common questions
Helpdesk auto-replies and macros handle the questions with one correct answer, like your return window or how to start an exchange.
Someone hears about each return
A Shopify Flow workflow can start when a return is requested, tag the order, and email the person who handles returns.
Before paying for anything new, set these up properly: write the return policy as rules, build macros for the top questions, and route tickets by topic. That alone can remove a lot of manual work, and it costs setup time rather than a build.
When the work crosses tools and needs glue
The trouble usually isn't the helpdesk or the returns app. It's the gaps between them. A return request lives in one tool, the order in Shopify, the package scan at the 3PL, and the follow-up email in Klaviyo. Each tool automates its own piece well. Nobody automates the handoffs, so a person does. Typical examples:
Refund only after the 3PL confirms receipt
The return is approved in one tool, the package is scanned at the 3PL, and the refund lives in Shopify. Glue connects the three and holds the refund until the receipt is in.
Order-status answers that need two systems
The customer asks in the helpdesk, the order is in Shopify, and the real shipping status is in ShipStation or the 3PL portal. Someone is copying between tabs.
Refunds above a limit wait for approval
Small, in-policy refunds are prepared and queued; anything over the limit you set, or outside policy, waits for a named person to approve.
Returns data reaching email and reporting
Return reasons and exchange outcomes need to reach Klaviyo segments or the weekly report, not stay in a returns app. See eCommerce reporting.
Exceptions with the full story attached
Damaged items, wrong items, and repeat returners land in one queue with the ticket, order history, and 3PL notes together, instead of four tabs.
What these share: facts spread across several tools, a check on whether they agree, and a decision about money. That is where custom glue earns its cost, as long as it includes an approval step. Automation that can refund without a person in the loop is the part founders worry about most, and with good reason.
The decision table
Score the returns or support task you have in mind against each row. If most answers fall in the first column, configure the app or helpdesk you have. If two or more fall in the second, custom glue is worth scoping.
| Factor | A returns app or helpdesk rules are usually enough | Custom glue with approval is worth scoping |
|---|---|---|
| Tools involved | One returns app or one helpdesk, plus Shopify. | Helpdesk, Shopify, the 3PL, and email all hold part of the answer. |
| Where the data lives | Everything the agent needs is visible in one screen. | Agents copy order numbers, tracking, and receipts between tabs. |
| Refund rules | Your policy is simple and the app's built-in rules match it. | Refunds depend on a 3PL receipt, an amount limit, or a check across tools. |
| Approval | Approving each request inside the app is enough. | Some cases need a named approver, a limit, and a record of who said yes. |
| Exceptions | Rare, and one person handles them without digging. | Common enough that gathering context is most of the work. |
| Cost of a mistake | A wrong tag or a late reply is easy to fix. | A refund sent twice, or before the item came back, is real money. |
How a refund approval flow works
Here is the shape of a refund flow that crosses the helpdesk, the store, and the 3PL, with a person in charge of the money. The steps are the same whichever tools you run.
- A return or refund request arrives. From the helpdesk, a returns app, or your store's return request page.
- The system gathers the facts. It pulls the order from Shopify, the ticket history from the helpdesk, and the return status from the 3PL.
- It checks against your written policy. Within the return window? Eligible item? Received at the warehouse? Amount under your limit?
- Clear cases are prepared; others go to a person. In-policy cases are packaged and ready. Anything over the limit, outside policy, or unclear goes to a review queue with the reason flagged.
- A named person approves. They see the order, the evidence, and the suggested action in one place, and approve, edit, or decline. You decide later whether small cases still need a click.
- The tools are updated and the run is logged. The refund or exchange is issued, the ticket is updated, the customer gets a reply, and every step is logged, with an alert if any step fails.
Two details matter more than the rest. First, the approval limit and the policy checks are written down by you, not guessed by the system. Second, the system shows its work: the approver sees why a case was flagged, not just that it was.
What stays human, and why
Automation is good at consistency and speed on the predictable parts. It is poor at reading a frustrated customer, weighing a loyal buyer's history, or deciding that this one time the policy bends. Keep those with your team, and give them the context so they decide faster.
A practical way to set the line: anything that sends money, changes an order, or goes to an upset customer starts as a draft that a person approves. Move a category to fully automatic only after you have reviewed enough of its drafts to trust it, and keep a log so you can check later.
Checklist before you automate returns or support
- Your return and refund policy is written as rules: window, eligible items, condition, who pays shipping.
- You know your top ticket topics and which ones have one correct answer.
- Every tool the workflow touches is listed: helpdesk, store, returns app, 3PL, email.
- The helpdesk's built-in rules, macros, and routing are set up and in use.
- A refund approval limit is set, and a named person approves anything above it.
- Exceptions have a home: one queue, one owner, with the context attached.
- Automatic customer replies start as drafts until you trust them.
- Every run is logged, and someone is alerted when a step fails.
- Someone on your team owns the workflow and checks it after policy changes.
Where Cadence Ridge fits
We build the glue between the helpdesk, the store, the 3PL, and email when the tools you have don't reach far enough. It is designed to run in accounts your brand owns, keep a person in charge of refunds and sensitive replies, and log every run. Each engagement starts with one workflow, scoped in writing.
See customer support automation for the support systems we build, and the pricing page for scope and starting prices. To see what returns handling costs you per year, try the ops cost calculator. If you're deciding whether Shopify Flow covers a workflow, read Shopify Flow vs. custom automation. If a vendor controls your helpdesk or store setup, start with the vendor exit checklist.