Gorgias Alternative for Shopify: How to Evaluate One

Support tooling is two products wearing one name, and stores rarely need both at the same standard.

The queue job is about inbound volume: messages landing in several places, answered twice or not at all, with no record of what was promised last time. What you are buying is one list, a way to split it between people, and the customer's order history beside the conversation so the answer takes one screen instead of three.

The reach job is about who you can talk to at all. It is invisible in every demo, because a demo always has a test contact in it. In practice it is the number that caps everything downstream: a tool that answers beautifully on a channel holding 40% of your customers is worth less than a plain one sitting where 85% of them are. If you are unsure which of the two your week is actually made of, which parts of a support platform a small store needs separates them properly.

Keaz sits on the reach side of that line and has an inbox rather than being one: labels, filters, a visible reply-window clock and Shopify order history next to the chat — what the platform does is the honest scope. It is a shared inbox for a Shopify store, not a general-purpose helpdesk across six channels. Worth checking against your own week before you shortlist anything.

Feature lists rot, and every vendor writes their own. These five are answerable from a trial account in an afternoon and still matter a year later.

1. How many of your customers can it actually reach? Ask for the percentage, not the channel list. Count the customers who bought in the last twelve months, then count how many of them you hold a usable, consented contact for on the channel the tool sends on. That single fraction sets the ceiling on every other number in the evaluation.
2. Does it read your shop, or only sit next to it? The useful test: can the person replying see the last order, the order count and what that customer has spent, without opening a second tab? In Keaz that is the contact details panel beside the conversation, and it is the difference between answering “where is my order” in ten seconds and in two minutes.
3. Can it grow the list, or only message it? Reach is not a fixed quantity, which is why question one is worth asking twice — once now and once in ninety days. Shopify lets you set the customer contact method to phone number or email, and documents that requiring both at once is not supported on standard plans (Shopify Help Center, retrieved September 2026), so how you configure checkout decides what you can collect. Stores that encourage the phone number at checkout move from roughly 22% capture to 65–70%.
4. Can you model the outcome before you commit? Any vendor can show you a case study. Ask for a number computed on your shop instead: the Forecast runs on your own Shopify data, and the per-industry ranges show what the first 90 days look like in your vertical.
5. Who operates it every Tuesday? Setup is a one-off; the weekly routine is not. If a tool needs someone to build macros, rules or campaigns by hand every week, that person is part of the price and belongs in the comparison next to the licence fee.

Switching support tools is the one migration your customers notice immediately, so the test has to run without a gap in coverage. Four steps, in this order.

1. Pick one narrow slice of real traffic. One channel, or one question type — order status is the usual choice because it is high volume and easy to grade. Do not test on internal messages; they never behave like customers.
2. Run both tools in parallel for one full billing cycle. A week flatters everything. A full cycle catches a promotion, a delivery wave and at least one bad Monday, which is where tooling either holds or does not.
3. Grade on time-to-first-useful-answer, not on time-to-first-reply. An instant acknowledgement that resolves nothing is a metric improving while the customer experience does not. Measure how long it takes to say something the customer can act on.
4. Test the exit before you commit to the entrance. Run an export on the trial account and open the file. A tool you cannot leave is a tool you will be negotiating with rather than buying from.

If part of what you are replacing is several people sharing one number, run that specifically — how small teams share one WhatsApp number covers what changes when a queue becomes shared, and the Inbox overview shows the open-and-done split most small teams end up using.

This is the part that gets discovered after the contract is signed, and it splits into two questions that are usually confused with each other.

The data question. Conversation history is customer data you are responsible for, and it is rarely portable by accident. The GDPR gives data subjects the right to receive their personal data in a structured, commonly used and machine-readable format (Regulation (EU) 2016/679, 2016), and you carry that obligation regardless of which vendor is holding the archive. Export it while the account is still live, check that the file actually opens, and keep it — a support archive is also your record of what you promised someone.

The consent question. Permission does not migrate with a contact list. A phone number that arrived in your old tool through a support conversation has not, by that fact, agreed to receive marketing from a new one. Keaz records how much consent each contact has given as an opt-in level and lets you set a minimum level as a filter on every marketing flow, so contacts you cannot vouch for never enter one. For anyone dormant beyond twelve months, Smart Activation asks again rather than sending, and it is capped at twice a year on purpose.

Treat those as two separate items on the migration checklist. Stores that merge them tend to move the data cleanly and then send their first campaign to people who never agreed to one.

Roughly in order of how often they cost a store a quarter.

• Comparing feature lists before naming the job. Two products in different categories both list “automation” and “inbox”. The comparison means nothing until you have written down which problem you are paying to remove.
• Optimising the queue when the problem is reach. A tidier queue answers the people who already found you faster. It does not add anybody. If your complaint is that a channel earns nothing, better routing is not the fix.
• Testing on a week instead of a cycle. Every tool survives a quiet week. Run the trial across a promotion and a delivery wave or you have graded the demo, not the product.
• Migrating contacts and assuming consent came with them. The data moves; the permission does not follow. Re-check opt-in status before the first send, not after the first complaint.
• Cancelling the old tool before the export opens. “We can export” and “the export is readable” are different claims. Verify the second one while you still have an account.
• Deciding on price per seat alone. Seats are the visible cost. The weekly hour someone spends maintaining rules, and the customers a channel cannot reach, are the expensive ones and neither appears on the invoice.