Gorgias AI Double-Billing Explained (Resolution + Ticket) vs Per-Action AI

The two things Gorgias meters, and where they overlap

Gorgias bills on usage rather than seats, which is a sensible model on its own. But it meters two different things, and the confusion starts where they overlap.

The first meter is the billable helpdesk ticket. Every plan bundles a quota of tickets; go over it and each extra ticket costs roughly $0.40 on Starter and Basic, $0.36 on Pro and Advanced, and $0.32 on Enterprise, per the pricing breakdowns compiled by Chatarmin and others. That's the classic help-desk unit.

The second meter is the AI resolution. When the AI Agent fully handles a conversation, Gorgias charges an automation fee: $1.00 per resolution on a monthly contract, $0.90 on annual, with interactions beyond your bundled AI allowance running $1.50 each. Those figures are consistent across independent breakdowns from myAskAI and Chatarmin, and reflect Gorgias's published AI Agent model.

Individually, both are defensible. The friction is that a single automated conversation can trip both meters at once.

When one conversation becomes two charges

Here's the rule, straight from Gorgias's own documentation. Per the How you're billed for using Gorgias help article, an automation fee and a ticket fee apply to the same ticket only when the ticket is entirely resolved by AI Agent, without human intervention. In other words, the cleanest, most fully-automated cases — the ones you'd think are cheapest — are exactly the ones that get billed twice.

There's an important qualifier that softens it, and it's only fair to state it plainly. Gorgias uses a 72-hour rule: an interaction only counts as "automated" if the customer does not require a human agent within 72 hours. If a teammate jumps in during that window, the conversation is billed as a ticket only — no separate resolution fee. So the double charge lands specifically on the conversations the AI closed cleanly and nobody reopened.

For a WISMO-heavy ecommerce inbox, that distinction matters enormously, because WISMO is precisely the category the AI resolves end-to-end without a human ever looking. Play it out on the numbers myAskAI models: 10,000 tickets a month, 50% resolved by AI, on an annual contract. The helpdesk plan runs roughly $2,550, and the 5,000 AI resolutions at $0.90 add about $4,500 — so the automation you switched on to save money more than doubles the bill, to around $7,050/month. One merchant myAskAI cites was on the Advanced plan at $13.5k/year and, after enabling AI, saw roughly $14,000 in resolution charges stacked on top.

None of this means Gorgias AI is a bad product. It means you have to model the combined cost — tickets and resolutions together — not read the headline plan price and assume you're covered.

Credit where it's due: what native Gorgias Automate and AI do well

It would be dishonest to frame this as "Gorgias bad." For a brand that lives inside Gorgias, the native automation is genuinely strong, and for a lot of teams it's the right first move.

• It's already there and already connected. Automate and the AI Agent read your Shopify order data, your macros, and your help center out of the box — no integration project. If you want a fuller tour of the native stack, our complete guide to Gorgias Automate and AI and the Gorgias AI Agent explained walk through it.
• It's tuned for ecommerce. WISMO, order tracking, and simple returns are the AI Agent's home turf, and it handles them with almost no setup.
• Outcome-priced automation is legible to finance. "We paid $0.90 and the ticket closed" is an easy sentence to defend in a budget review.

Where it stops is worth naming just as plainly. First, resolution-based billing means you're paying for an outcome definition Gorgias controls — and the definition bills you twice on the exact conversations you automated most cleanly. Second, "resolution" flattens a lot of nuance: a bot that deflects with an FAQ link, a bot that actually does something (issues the return, edits the order), and a bot that just replies convincingly all count the same to the meter, even though their value to the customer is wildly different. That gap — deflection versus true resolution versus automation — is the heart of our take on AI agents for customer service.

The per-action alternative: pay for the work, not the label

Here's where a different unit changes the math. Instead of a flat fee per "resolution," a per-action credit model charges for the discrete AI actions a conversation actually consumes — a message drafted, a tool called, an order looked up. You pay for work done, not for a binary outcome that may or may not have required much work at all.

Macha is an AI agent layer built this way. To be clear about what it is: Macha runs on top of the Gorgias you already use through a live native connector — it is not a Gorgias replacement and it doesn't touch your helpdesk of record. It reads and writes the same Gorgias tickets, and you keep Gorgias for everything Gorgias is good at. What changes is the meter. Credits in Macha are consumed per AI action, so a simple WISMO reply that took one grounded message costs far less than a multi-step return that fetched an order, checked a policy, and posted a resolution — and neither one is quietly billed a second time as a separate helpdesk ticket you already pay Gorgias for.

The screenshot below is from the Macha demo org: an Escalation Manager agent whose config surfaces its cost right on the action — 3 credits per message — so the price is attached to the unit of work, not to a resolution label.

The practical difference on an ecommerce inbox:

The point isn't that per-action is universally cheaper — a conversation that genuinely takes ten actions can cost more than a flat fee, and that's honest. The point is that the price tracks the work, and you're never billed twice for a single automated conversation. For a high-volume, low-complexity WISMO inbox, that asymmetry usually lands in your favor.

Setting it up without ripping anything out

Because Macha is a layer, adoption is additive, not a migration.

1. Keep Gorgias as your system of record. Your inbox, macros, views, and Shopify data stay exactly where they are.
2. Connect Macha to Gorgias via the native connector — the step-by-step is in connect Gorgias to Macha for automated ticket replies.
3. Give the agent real tools. Point a custom tool at your order or subscription API so the agent can fetch a tracking status or start a return itself, rather than just linking to an FAQ.
4. Decide the split. Let per-action agents carry the high-volume, repetitive work (WISMO, tracking, simple subscription changes), and route the genuinely complex cases to humans — the same handoff Gorgias's 72-hour rule already rewards.

If you're still comparing the native route head-on, our Gorgias pricing explained breakdown lays out the plan tiers and where the AI fees attach.

FAQ

Does Gorgias really charge twice for an AI-resolved conversation? In one specific case, yes. Per Gorgias's billing documentation, an automation (resolution) fee and a helpdesk ticket fee apply to the same ticket only when the AI Agent resolves it entirely, with no human involvement. If a human replies within 72 hours, it's billed as a ticket only — no separate resolution fee.

How much is a Gorgias AI resolution? Independent breakdowns and Gorgias's model put it at about $1.00 per resolution on monthly billing and $0.90 on annual, with interactions beyond your bundled AI allowance at $1.50 each. Confirm the current figures against your own Billing & usage page, as pricing changes.

What does "per-action" billing mean instead? Rather than a flat fee per resolved conversation, you're charged for the discrete AI actions a conversation consumes — a message sent, a tool called, an order looked up. A one-message WISMO reply costs less than a multi-step return, and a single automated conversation isn't billed twice.

Does Macha replace Gorgias? No. Macha is an AI agent layer that runs on top of Gorgias through a live native connector. Gorgias stays your help desk and system of record; Macha reads and writes the same tickets and adds a per-action AI agent on top.

Which model is cheaper for my store? It depends on your mix. High-volume, low-complexity inboxes (lots of one-message WISMO) tend to favor per-action credits, since you avoid paying a full resolution fee — plus a ticket fee — on trivial conversations. Model your real ticket and resolution volumes together before deciding.

Want to see per-action economics against your own Gorgias volume? Start a free trial of Macha and connect it to your Gorgias in minutes.