Gorgias Setup for Shopify: The Step Most Teams Skip

What Gorgias actually changes for a Shopify support team

Gorgias is a customer service platform built for ecommerce, and for a Shopify store its main job is pulling order, shipment and customer data into a support ticket so an agent — or its Ask AI feature — never has to leave the inbox to check an order status. That’s the pitch every vendor page makes. What they don’t cover is the setup order: get the rules and the AI hand-off threshold wrong, and Gorgias for Shopify produces faster wrong answers instead of faster right ones.

This guide covers stores that already have ticket volume worth automating — roughly $3M to $30M in revenue, on Shopify Plus or a comparable paid platform, with a support team of at least two people. If you’re running a single shared inbox and a few hundred orders a month, Gorgias’s rule engine and Ask AI tier are more machinery than your ticket volume needs; a simpler helpdesk with canned replies will do the same job for less setup time.

Before you connect anything: what you need

Gather three things before you touch the Gorgias admin. First, admin access to your Shopify store — the integration needs write permissions to update orders and process returns, not just read access. Second, a list of your actual support scenarios, written from real tickets over the last month, not from your FAQ page. Third, a decision on what Ask AI is allowed to do without a human: reply, tag, or also refund and cancel. Make that decision before setup, because Gorgias ships Ask AI with broader action permissions on by default than most $3M-plus brands want live on day one.

Step 1: Connect the Shopify channel correctly

In Gorgias, go to Settings, then Integrations, and add Shopify before adding any other channel — email, live chat, Instagram. Order matters here because Gorgias uses the first ecommerce integration you connect to set the default customer-matching logic for every ticket after it, matching by email and then by order number. Connect a second Shopify store to the same Gorgias account later, and tickets from that store won’t inherit the same matching rule automatically — you’ll need to set it per integration in the Shopify app’s settings panel.

Once connected, check that the integration is pulling order status, not just order existence. Open a test ticket from a real customer email address and confirm the sidebar shows fulfilment status, tracking, and line items, not just an order number. A connection that shows the order but not its fulfilment state usually means the store’s fulfilment app isn’t writing status back to Shopify’s core order object, and Gorgias can only display what Shopify itself has recorded.

Step 2: Set up rules before macros

Most teams build macros first because macros are visible and satisfying — you write a reply, you see it. Rules are invisible until they misfire, which is exactly why they need to go first. A rule in Gorgias runs automatically against every incoming ticket matching its conditions: tag by keyword, route by channel, autorespond by order status.

Build rules in this order: tagging rules first (order status, product line, sentiment), routing rules second (which team or agent sees the ticket), autoresponder rules last, because an autoresponder that runs before tagging can close a ticket a later rule would have handled differently. Putting an autoresponder rule before its tagging rule means the autoresponder fires on a ticket that a later rule would have routed differently, and the customer gets a reply meant for a different scenario.

A concrete example: a rule that autoresponds “your order has shipped” to any ticket containing the word “shipping” will also fire on a ticket that says “my shipping address is wrong,” because the keyword match doesn’t check intent. Write the condition as an exact-phrase or regex match against “where is my order” or “shipping status,” not a bare keyword, and exclude tickets already tagged “address change” by an earlier rule.

Step 3: Build macros for the SOP, not the FAQ

A macro should exist because a real agent typed a real answer more than once last month, not because a page exists on your website. Teams that build a macro for every FAQ section end up with a library of thirty or more near-identical replies, and both agents and Ask AI struggle to pick the right one when three macros all cover “where is my order” with slightly different wording.

Build one macro per operational scenario: a delayed order past its estimated window, a wrong item received, a return request outside the stated window, a duplicate charge. Each macro should include the specific action the agent takes — issue a label, escalate to a supervisor, apply a store credit code — not just the sentence the customer reads. That action line is what Ask AI reads to decide what it’s allowed to do, so a macro without one gives the AI a reply to send but no instruction on what happens next.

Keep macro variables tied to real Shopify fields — customer first name, order number, tracking link — pulled from the integration, not typed in by hand. A macro with a hand-typed placeholder like “[insert tracking here]” will occasionally go out unedited under ticket-volume pressure, and that’s the ticket that becomes a public review.

Step 4: Configure Ask AI hand-off settings

The Ask AI hand-off is the step most Shopify teams get wrong, and it’s a permissions setting, not a tone setting. Ask AI’s default configuration allows it to resolve tickets across a broad set of intents, including some refund and cancellation flows, without pausing for a human. That default is built for speed, and it’s the wrong default for a brand where a wrong refund costs more than the agent-minute it takes to check.

Go to Ask AI’s settings and restrict its autonomous actions to intents where a wrong answer is cheap to fix: order status checks, tracking lookups, standard return label issuance within policy. Route anything involving a refund outside policy, a damaged high-value item, or a customer who has already contacted support twice on the same order to a human queue instead of letting Ask AI resolve it. The setting is per-intent, so this isn’t an all-or-nothing switch — it’s a list you build scenario by scenario, the same list you used to write macros in Step 3.

Test this by sending Ask AI five real past tickets that your team resolved manually, including at least one refund-adjacent one, and checking what it would have done. If it would have refunded a case your team escalated, the intent boundary needs tightening before it goes live on real traffic.

Step 5: Route Instagram and email without double-answering

Once Shopify, email and any social channels are all connected, check for overlap in your routing rules. A common failure mode: a customer DMs on Instagram and emails support about the same order within an hour. Without a cross-channel dedupe rule, both tickets get answered independently, sometimes with contradictory information if one agent has more current order data than the other.

Gorgias’s shared inbox links tickets from the same customer profile when the contact details match, but it doesn’t auto-merge them unless a rule tells it to. Add a rule that tags a new ticket as a possible duplicate when an open ticket exists for the same customer within a set window — a few hours is usually enough to catch same-issue, different-channel contacts without flagging genuinely separate questions from a repeat customer.

The setting most teams get wrong

The Ask AI action-permission setting is worth checking today rather than at the end of a rollout. Teams treat Ask AI’s launch as a macro-writing exercise and skip past the permissions screen on its default, which is where the actual risk sits. A macro with clumsy wording produces an awkward reply; an AI with refund permission on a scenario it shouldn’t have produces a wrong refund, and that’s a support ticket that becomes an accounting problem.

Rule order trips up teams too: they write autoresponders first because they’re the most visible win, then add tagging rules afterward that never get a chance to run because the autoresponder already closed the ticket.

How to verify the setup is working

Pull a week of tickets after go-live and check three things: the rate of tickets reopened within 24 hours of an automated resolution, the rate of duplicate tickets across channels for the same order, and a manual review of every ticket Ask AI resolved with refund or cancellation permission. A rising reopen rate usually traces back to a rule or macro that resolves a ticket before the actual issue is fixed — the customer gets a reply that looks complete but doesn’t address what they asked.

There’s no universal benchmark reopen rate to hit, because it depends on your product return complexity and order volume; treat any fixed percentage you see quoted online as unverified until you’ve measured your own baseline before and after the Gorgias rollout.

Who this setup is not for

If your support volume is low enough that one person handles it from a personal inbox, the rule-building and Ask AI permission work in this guide is overhead you don’t need yet — a simpler shared inbox with saved replies covers the same ground with a fraction of the setup time. This guide is written for a team already choosing between agents and automation, not one deciding whether to get a helpdesk at all.

Gorgias setup for a Shopify store is a customer service automation problem before it’s a support-software problem: the rules, macros and Ask AI permissions above only work if they’re built around your actual ticket scenarios and reviewed as volume changes, not configured once and left. If that ongoing rule and permission work is more than your team has capacity for, that’s the gap customer service automation covers, and it’s the same operational discipline that matters most for scaling brands whose ticket volume is growing faster than their support headcount.