Zendesk Pricing Plans: What You'll Really Pay

Zendesk pricing plans list one number per tier — $19, $55, $115 — and that number is not what a Shopify support team above the $3M floor actually pays by the time Copilot, Contact Center and an unpublished AI overage rate are added to the invoice. The headline prices are real and Zendesk publishes them plainly. What it does not publish, anywhere on its own pricing page, is the rate it charges when its AI agent resolves more tickets than your plan includes for free — and for a team running any real ticket volume, that single unpublished line usually outweighs every seat fee combined.

This piece works through what Zendesk actually charges, tier by tier and add-on by add-on, then builds one combined monthly bill for a mid-size ecommerce support team so the arithmetic is visible instead of implied.

What do Zendesk pricing plans actually include and cost?

Zendesk sells four Suite and Support tiers, all priced per agent per month on annual billing, with month-to-month available at what the pricing page itself calls “a higher effective monthly rate” — without stating what that rate is.

Two things about Zendesk’s published tiers are easy to miss if you read only the price column. First, Suite Enterprise has no public number to compare — it sits behind a sales conversation, so any “Zendesk costs $19 to $115” claim, including the one at the top of this article, silently excludes the tier Zendesk expects its larger customers to buy. Second, the annual price is the only price Zendesk shows in public. A team that plans to pay monthly, for cash-flow or contract-flexibility reasons, cannot get an actual figure from the pricing page at all — only the statement that it costs more.

What add-ons turn a $115 seat into a bigger bill?

Three separately priced add-ons sit on top of the Suite tiers, and each is billed per agent per month regardless of which base plan the agent is on.

Stack all three onto a Suite Professional seat and the per-agent price moves from $115 to $298 before a single ticket is answered — before, importantly, the AI agent resolves anything at all. Suite Enterprise + Copilot folds Copilot into its quote, which is one reason a straight Enterprise-versus-Professional comparison on seat price alone understates what Enterprise actually includes.

None of the three add-ons is optional in practice for a team that wants what most ecommerce brands actually need: AI-assisted answers (Copilot), a phone channel for high-value or urgent orders (Contact Center) and scheduling for a support team that is not staffed 24 hours a day (Workforce Engagement Bundle). A brand evaluating “Zendesk pricing plans” against a $115 headline is evaluating a number almost nobody on Suite Professional actually pays.

What does Zendesk charge when its AI agent resolves more tickets than the plan includes?

It charges a per-resolution overage fee, and the rate is not published anywhere on Zendesk’s own pricing page — that absence is the single largest hidden line item in this entire pricing structure. Zendesk defines “Automated Resolutions” as the billing unit for AI agent usage: a Verified Resolution, where the AI closes a ticket and its own model confirms the closure resolved the issue, is what gets counted and billed. Every plan includes some number of free resolutions tied to seat count, and the allowance is small enough, relative to real support volume, that a team running its AI agent seriously will exceed it inside the first busy week.

What happens past that allowance is where the published price stops answering questions. Zendesk’s pricing page defines the billing unit and states that AI agent usage is “based on the successful outcomes they deliver,” but it does not attach a dollar figure to a resolution anywhere a prospective customer can read before talking to sales. Figures in the $1.50-to-$2.00-per-resolution range circulate widely across third-party pricing guides, but every one of those figures traces back to customer-reported invoices and contract teardowns rather than a Zendesk rate card, and several such guides repeat the same range without an independent measurement behind it — which is the pattern that makes a number worth treating as unconfirmed rather than quoting as fact. The actual overage rate for your contract is — metric to confirm, and the only way to close that gap is to ask your account rep for the committed-pack rate and the pay-as-you-go rate in writing, at the tier you are actually buying, before you sign.

What does a full Zendesk bill look like for a $3M-$30M Shopify support team?

It runs well past the advertised per-seat number once seats, add-ons and AI overage are combined into one bill — the combination Zendesk’s own pricing page never runs for a prospective buyer, and the one the most detailed published third-party pricing breakdown, from Macha, explicitly declines to build, saying it prices Copilot and the resolution meter but stops short of folding in Contact Center or the Workforce Engagement Bundle. This worked example uses invented figures for a plausible mid-size Shopify support team rather than a specific customer’s real invoice, and the arithmetic — not the inputs — is the part worth trusting.

Assume 12 agents on Suite Professional, all 12 with Copilot, 3 of the 12 also covered by Contact Center for phone escalations, and Workforce Engagement Bundle across the full team. Assume the AI agent closes 3,000 tickets in a month against a free allowance illustratively set at 10 resolutions per agent — 120 free resolutions total — with the remaining 2,880 billed at an illustrative $2 per resolution, a round number chosen for the example and not a quote from Zendesk, whose actual rate is unpublished. Every input in this paragraph — the agent count, the ticket volume, the free allowance and the per-resolution rate — is invented for illustration rather than measured, and none of it should be quoted back as Zendesk’s real numbers; only the method of adding seats, add-ons and overage into one bill is the point.

That total works out to roughly $716 in effective cost per agent per month — about six times the $115 headline seat price the pricing page shows first — and the overage line by itself, $5,760, is larger than every seat and add-on charge combined ($2,829). That is the shape of the hidden cost: it is not that any single add-on is expensive, it is that the free AI allowance is sized to seat count while real ticket volume is sized to order volume, and the two scale at completely different rates once a brand is doing enough business to need AI deflection in the first place.

How does Zendesk’s AI automation compare with what Gorgias reports?

It compares on a different basis entirely, because Zendesk’s public figures are prices and Gorgias’s public figures are outcomes. Gorgias, a Shopify-native helpdesk that competes directly with Zendesk in ecommerce support, markets its AI Agent as able to “automate 60% of support” and cites a case study — Orthofeet, vendor-reported — reaching a 56% automation rate in under two months. Both figures come from Gorgias’s own marketing and case-study material, so they belong in this article labelled exactly that way: vendor-reported, not independently measured, and not directly comparable to anything Zendesk publishes about its own AI agent’s resolution rate, because Zendesk has not put a matching outcome figure in public.

What the two vendors do share is the billing mechanic: Gorgias’s AI Agent is also sold on a performance-based, fee-per-resolution model, which means a brand switching helpdesks to escape Zendesk’s opaque overage rate is not automatically escaping usage-based AI billing altogether — it is trading one vendor’s unpublished rate for another vendor’s unpublished rate. The genuine difference, if there is one worth paying for, sits in the automation rate each tool actually reaches on your own ticket mix, and neither vendor’s marketing page can answer that; only your own ticket data run through a trial can.

When does Zendesk stop being worth its price?

It stops being worth the price at the point where the AI overage line becomes the least predictable number in your support budget — not at any fixed dollar threshold, because that point depends on your ticket volume and your AI resolution rate, both of which are specific to your product and your customers. A twelve-agent team whose AI agent resolves a few hundred tickets a month barely touches its free allowance; the same team resolving a few thousand tickets pays an overage bill several times larger than every seat and add-on charge combined, which is the shape of the mechanism regardless of the exact agent count or ticket volume involved. Zendesk’s own pricing page gives no way to see which side of that line your account will land on before you sign, because the rate that decides it is not published.

What should you actually ask Zendesk before you sign?

Ask for four numbers in writing, because none of the four appears on the public pricing page, and a verbal figure from a sales call is not a contract term.

The committed-pack rate and the pay-as-you-go rate for Automated Resolution overage, quoted for the specific tier you are buying rather than as a general range — a twelve-agent team’s overage line can run several times larger than its entire seat-and-add-on bill combined, so this single number decides more of your total cost than any seat price does. The exact free-resolution allowance per agent per month for your tier, since that figure has moved during 2026 and a stale number from a blog post will understate your real exposure. The month-to-month price for your tier, if annual billing is not an option for your team, since Zendesk states only that it costs more without stating what more means. And whether Copilot, Contact Center and the Workforce Engagement Bundle can be added or removed mid-contract as your agent count changes, or whether each is locked to the term you sign.

A support budget built on the $19-to-$115 headline and silent on all four of those numbers is a budget built on the part of Zendesk’s pricing that was designed to be quoted, not the part that determines what you actually pay.

Swapping vendors is not automatically the fix, since usage-based AI pricing is common across ecommerce helpdesks rather than unique to Zendesk. The more durable alternative is building the automation layer against your own ticket taxonomy and your own escalation rules, so the cost scales with what you actually built rather than with a per-resolution meter set by a vendor. A support stack whose single largest line item is a metered AI rate the vendor will not put in writing is not a procurement inconvenience; it is what happens when ticket deflection is bought off a price list instead of built against your own queue. Pointerflow’s customer service automation work starts from a ticket audit of your own six most common question types and builds the automation as infrastructure you own, so the monthly cost is a fixed build rather than a meter that moves with every order surge.