How to calculate cost per support ticket
To calculate cost per support ticket, add your fully loaded labor cost, your tool and software cost, and your overhead for a given month, then divide by the number of tickets resolved in that month. That is the whole formula. The work is in getting each input right, because most stores either use headline salary instead of fully loaded labor, or count tickets received instead of tickets resolved, and both mistakes push the number in a direction that hides the real cost.
The formula in one line
Cost per resolution = (labor + tools + overhead) ÷ tickets resolved.
Skip industry benchmarks. A "$3 to $5 per ticket" figure floating around a blog post assumes a labor rate, a tool stack, and a ticket mix that has nothing to do with your store. A brand selling $40 candles with three-word tickets ("where's my order") has a completely different cost structure than a brand selling $2,000 furniture with tickets that involve freight claims and photos. Run your own numbers instead of borrowing someone else's.
Step 1: add up fully loaded labor cost
Start with wages, not salary. If an agent earns $22 an hour and works 160 hours a month, that's $3,520 in wages. Add payroll tax, workers' comp, and benefits, which typically run 15 to 25% on top of wages depending on your state and whether you offer health coverage. At 20%, that agent actually costs you $4,224 a month, not $3,520.
Do this for everyone who touches a ticket, not just the person with "support" in their job title. If your operations manager spends six hours a week triaging escalations, put a real dollar figure on those hours and add it in. Support ticket cost calculations that only count the front-line agent's paycheck consistently understate the true cost per resolution, sometimes by 20 to 30%, because they leave out everyone above the agent who still touches tickets.
Step 2: add tool and software costs, including the per-seat trap
Add up every subscription that touches the ticket flow: helpdesk, live chat, returns portal, order lookup app, phone system. Most helpdesk platforms price per agent seat, so a five-agent team on a $75-per-seat plan pays $375 a month regardless of whether they handled 2,000 tickets or 6,000.
This is the trap. Per-seat pricing scales with headcount, not with ticket volume, so as you hire to keep up with growth, your tool bill climbs in lockstep even if automation could have absorbed half those tickets without a new hire. If you're evaluating alternatives, see how a Gorgias alternative or a Zendesk alternative changes this line item, because a system you own instead of rent removes the per-seat scaling entirely.
Step 3: add overhead, training, QA, and management time
Overhead is the cost of everything around the ticket, not the reply itself. Three things belong here, and you don't need more than three:
- Onboarding and training: if it takes a new agent two weeks at partial productivity to ramp, put a dollar figure on that lost output and spread it across the months you expect to keep that agent.
- Quality review: if someone spends four hours a week listening to calls or spot-checking replies, that's real payroll cost tied to support, even though it doesn't touch a single ticket directly.
- Management time: if a manager spends 20% of their week on the support queue, scheduling, escalations, reporting, that 20% of their loaded cost belongs in this calculation.
Don't try to allocate rent, electricity, or your accounting software here. Cost per resolution gets less accurate the more indirect line items you force into it. Three categories, applied consistently month to month, beats a spreadsheet with forty rows nobody updates.
Step 4: divide by tickets resolved, not tickets received
This is where most calculations quietly go wrong. Tickets received counts every message that hits the inbox, including duplicates, auto-replies bouncing back, and the same customer following up three times on one issue. Tickets resolved counts each customer problem once, at the point it's actually closed.
If a customer emails, gets an auto-reply, replies again in frustration, and then gets a real answer, that's one resolved ticket, not three received tickets counted separately. Count it as three and your cost per ticket looks artificially low, because you're dividing the same labor cost by an inflated denominator. Count reopens as new tickets and you get the opposite distortion. A resolved problem that reopens a week later isn't a second sale of the same labor. It's a sign the first resolution didn't hold.
Worked example: a $250k a month ecommerce store
Take a store doing $250,000 a month in revenue with three support agents and a part-time manager.
Labor: three agents at $3,800 fully loaded each equals $11,400. The manager spends 15 hours a week on support out of a $5,500 monthly loaded salary, roughly 37.5% of their time, adding $2,062. Total labor: $13,462.
Tools: a helpdesk at $79 per seat for four seats (three agents plus manager access) is $316, plus a returns app at $99 and an order-lookup integration at $49. Total tools: $464.
Overhead: quality review at three hours a week from a $25-an-hour lead adds $325. Training for one new hire that month, two weeks at half productivity on a $3,800 loaded cost, adds roughly $475. Total overhead: $800.
Total monthly support cost: $13,462 + $464 + $800 = $14,726.
If that store resolved 2,600 tickets that month, cost per resolution is $14,726 ÷ 2,600 = $5.66 per ticket. Most of that $464 tools line is the per-seat helpdesk charge, the part of the bill that climbs with headcount rather than with ticket volume. Run your own store's figures through the cost calculator to see where you land.
Why the number keeps climbing every month
Customer service costs rising every month usually isn't one dramatic event. It's three small mechanisms compounding.
Order volume grows faster than headcount because hiring lags revenue by design. Nobody hires ahead of a sales spike. Average handle time creeps up as product lines expand, since more SKUs mean more edge cases per ticket and agents spend longer looking things up. And every time you add an agent, your per-seat tool cost adds another line item, so tool spend grows in steps even when ticket volume grows smoothly.
None of this shows up as a single line item getting more expensive. It shows up as cost per resolution drifting upward quarter over quarter even though nobody changed a price. If you want to lower average handle time without new hires, the fix usually isn't training. It's removing the lookups that eat the time in the first place: order status, tracking, and return eligibility. Take those away and agents spend their minutes on tickets that actually need a human judgment call.
Where automation actually moves the number
Automation doesn't reduce ticket volume to zero, and any claim that it does isn't one we'd make without showing the math. What it does is remove the categories of ticket that don't need a person: order status and tracking lookups, standard return and exchange approvals, and first-pass fraud review on orders that match known patterns.
Those three categories typically make up a large share of inbound volume for ecommerce stores, because they're the tickets a customer sends when they can't self-serve the answer, not because the issue is complicated. Automating them lowers your labor line directly, since fewer agent-hours go to tickets that never needed a human. It also caps your overhead growth, because you're not onboarding a new agent every time order volume ticks up. Tool spend doesn't go to zero. You still need infrastructure to run the automation, but it stops scaling per seat, which is the mechanism that was inflating cost per ticket in the first place. See how this works in practice on our ecommerce support automation page.
Plug your own numbers in
The mechanics don't change with store size. What changes is your labor rate, your tool stack, and how many of your tickets a person actually needs to touch, and those three numbers are yours to measure, not to borrow. Run them through the cost calculator and get your own number instead of a guess.
Common questions
Is there an industry benchmark for cost per support ticket?
Not a useful one. Labor rates, tool stacks, and ticket complexity differ too much between businesses. Calculate your own with the formula above instead.
Should cost per ticket include reopened or duplicate tickets?
Only count each ticket once at final resolution, otherwise reopens inflate your ticket count and understate your real cost per resolution.
How is cost per ticket different from average handle time?
Average handle time measures minutes spent per ticket. Cost per ticket converts that time, plus tools and overhead, into a dollar figure.
How much does a fractional COO cost compared to building a support system once?
A fractional COO is an ongoing monthly fee. A support system is a one-time build you own. The comparison depends on how long you'd keep paying for either.
Once you've run your own cost per resolution, the next question is what a system replacing that manual work would actually cost to build and whether it pays for itself faster than another quarter of rising ticket volume.
Find out what the manual work is costing you
One call, we map where the time goes and what a system would replace.