If you run a Shopify store doing more than 500 support conversations a month, Gorgias is quietly eating your margin. Per-ticket pricing made sense when AI agents could not resolve anything autonomously. That stopped being true. This is what the switch looks like, what to watch for, and what 47 operators told us about the move.
The math is brutal once you do it on a napkin. A store doing 2,000 conversations a month on Gorgias's Pro tier is paying roughly $1,000 to $1,800 depending on which add-ons you carry. That same volume on Arbyn is $149. Flat.
What changed is not just the price. It is what you actually buy at that price. The category split between a help desk you pay per ticket for and an AI agent that resolves the conversation for you used to be real. It has collapsed.
Why the math changed
Three things happened in the last 18 months. AI agents started resolving the long tail of customer questions reliably, not just the easy 20 percent. Shopify Functions opened up the checkout, which means the support tool can now act on cart, shipping, and payment data without API workarounds. And the cost of LLM inference dropped enough that a per-conversation business model became uncompetitive at any volume above hobbyist.
Together that means a flat-rate AI-native tool can do what a per-ticket help desk does, charge nothing per resolution, and still have a healthy margin. The legacy per-ticket model is now a defense of yesterday's gross profit, not a reflection of today's costs.
What 47 operators actually did during the switch
We interviewed every Shopify operator who told us they made the move from Gorgias in the last six months. The pattern is consistent enough to write down.
1. They ran both tools in parallel for two weeks
Almost everyone keeps Gorgias active while Arbyn comes online. The first week is shadow mode: Arbyn drafts every reply and routes it to your draft queue. You approve or edit. Nothing leaves your store unless you say so.
The second week is the gradual handover. Arbyn starts sending the high-confidence drafts automatically while still holding the low-confidence ones for human review. You watch the resolution rate, CSAT, and any escalations. If anything looks wrong, you flip the kill switch.
2. They imported macros and saved replies on day one
Your Gorgias macros, saved replies, and rule book are the closest thing you have to documented brand voice. We pull them automatically during onboarding so Arbyn has a starting library before it has read a single email. This shortcuts the calibration time by roughly a week.
3. They cancelled Gorgias on a Tuesday, not a Friday
The operators who switched without drama cancelled their previous tool mid-week. A Friday cancellation followed by a weekend support spike is a bad first impression. Tuesday gives you four business days to spot anything weird before the volume goes up.
Doing the cost math for your store
Before you switch, know the number. Open your Gorgias billing dashboard for the last three months. Add up what you actually paid, including overages. Then multiply by 12.
Now do the same math for the next 12 months at your projected growth rate. Most operators we spoke with had budgeted between 30 and 60 percent volume growth for the year. That means the next 12 months on Gorgias would have been higher than the last. The next 12 months on Arbyn is the same number whether your volume doubles or stays flat.
What you keep when you switch
- Your inbox. Arbyn connects to your existing email address. Your customers do not see a new sender.
- Your tone. The Voice Fingerprint feature reads your sent history and matches it. Operators report 91 percent of drafts approved without editing after the first 30 conversations.
- Your team setup. Unlimited seats are included. Move your whole team over, no per-seat pricing surprises.
- Your rules. Refund ceilings, escalation triggers, restricted actions on specific order tags. You set them up once and Arbyn never crosses them.
What you throw away
This is the part nobody writes about. Some of the things you depend on inside Gorgias do not exist inside Arbyn. Most of the time that is intentional. Sometimes it is just different.
- Ticket counts that grow with revenue. Per-ticket dashboards are a vestige of paying per-ticket. Arbyn tracks resolutions and revenue attribution. The numbers on the dashboard are different but they tell you what you actually care about.
- Macro management as a daily activity. Macros still exist but they are mostly imported on day one and then mostly ignored. The agent rewrites the response from your store data each time, so the macro is a hint, not a script.
- Per-agent assignments. Arbyn is the agent. Your humans handle escalations. The "who is on shift" routing logic from a per-seat help desk does not transfer because it is not needed.
What the first week actually looks like
The most common worry operators told us they had was: "what if it gets one wrong on day three and a customer screenshots it on Twitter." The honest answer is that shadow mode exists precisely so this does not happen. For the first 30 conversations, Arbyn drafts but never sends. You sign off on the calibration before it goes live.
After that, the kill switch and the confidence threshold are the two controls that matter. The threshold can be set conservatively at the start, which routes more drafts to human review and gradually moved up as you gain confidence. The kill switch pauses everything in one click.
Bottom line
If you are running a Shopify store doing more than 500 conversations a month on Gorgias, switching to Arbyn is a 9 to 12 minute install that pays back its first year in the first month. The risk during the switch is contained by shadow mode, kill switch, and 21 days free to test it on your real volume. If it does not work for your store, you cancel from your Shopify admin and your invoice for the month is zero.
The decision is not whether per-ticket pricing is sustainable. It is not. The decision is whether you want to be the operator who switches early and saves the difference for 18 months, or the operator who waits and lets the budget swing for nothing.