Your support cost per ticket is your total support cost for a period divided by the number of tickets you resolved in that period. Total support cost means more than agent salaries. It includes benefits and payroll taxes, the salaries of leads and managers, every tool in the support stack, and a share of office and admin overhead. Most teams quote a number that only counts front-line wages, which is why the figure they carry around is usually too low by a wide margin.
What goes into a real support cost per ticket
The formula itself is not complicated. The honesty is in the inputs. Total support cost, divided by resolved tickets, for the same period. HDI, a support-industry body, defines it the same way: "Cost per Ticket is the total monthly operating expense of a service desk divided by the monthly ticket volume." The word doing the work is total.
Benchmarking firm MetricNet lists the operating expense in five buckets: salaries and benefits for agents, salaries and benefits for indirect staff like team leads and trainers and QA, technology and telecom costs, facilities costs, and training and travel and office supplies.
Here is the same breakdown laid out as a checklist, with rough proportions for a typical small team. Treat the percentages as a rule of thumb, not a benchmark.
| Cost bucket | What it includes | Rough share of total | |---|---|---| | Front-line labour | Agent salaries, benefits, payroll taxes, paid time off | Usually the largest line | | Indirect labour | Team lead, support manager, QA, scheduling, the founder's hours on escalations | Often forgotten entirely | | Tooling | Helpdesk, knowledge base, phone or chat software, AI tools, reporting add-ons | A meaningful slice, easy to undercount | | Overhead | Office space, utilities, IT support, HR time, recruiting and onboarding | Small per ticket, real in aggregate | | Hidden rework | Repeat contacts, escalations, refunds caused by a bad answer | Rarely measured, often the worst offender |
The headline number most teams use is too low because it counts only the first row. A salary is not just the gross figure on the contract: employer payroll taxes and benefits add a substantial layer on top, and that layer differs by country. The manager who spends a third of her week on support is a third of a salary that belongs in the calculation. The recruiter fee and the ramp time for a new hire are real money.
How to measure it without fooling yourself
Pick a clean period. A full month is the usual choice because most cost data arrives monthly anyway.
Add up every cost from the table above for that month. For salaried staff, take the fully loaded cost, gross pay plus employer contributions and benefits, and prorate anyone who only spends part of their time on support. For tooling, total the monthly bills and divide annual contracts by twelve. For overhead, a simple per-head allocation is fine: total facilities and admin cost, divided by company headcount, times the number of support people.
Then count tickets. Decide upfront whether your denominator is tickets received or tickets resolved. Resolved is the more honest choice, because an unresolved ticket still cost you money and often costs you again later. Be consistent so the trend line means something.
Watch repeat contacts in particular. If a customer writes three times about one problem, that is three tickets in your denominator and three contacts' worth of cost in your numerator. A low cost per ticket built on a pile of repeat contacts is not a low cost. It is the same expensive problem, counted in smaller slices. This is the measurement honesty we argued for in our piece on deflection rate: the metric is only as good as the definition under it.
A worked example
The numbers below are an illustrative example, not measured data, sized to look like a real European SMB so the method is clear.
Picture a six-person support team. Four agents at a fully loaded cost of 4,000 euros a month each. One team lead at 5,500 euros. The founder spends a quarter of her time on escalations; her loaded cost is 8,000 euros, so 2,000 euros belongs here. Front-line and indirect labour together: 23,500 euros.
Tooling, helpdesk and knowledge base and reporting, comes to 900 euros a month. Overhead, office and IT and HR and recruiting allocated across the team, works out to roughly 1,800 euros. Total support cost for the month: 26,200 euros.
The team resolved 3,200 tickets that month. Cost per ticket: 26,200 divided by 3,200, about 8.19 euros.
The napkin version would have said something else. Counting only the four agents' salaries, 16,000 euros over 3,200 tickets, the team would have told itself 5 euros a ticket. The real figure is 64 percent higher, and every plan built on the 5-euro number was built on sand.
What levers actually bring support cost per ticket down
Once the real number is on the table, the work is finding the levers that move it without hollowing out the customer experience.
Cut repeat contacts first. This is the highest-value lever and the most ignored. A ticket solved correctly the first time is one ticket. The same ticket solved badly becomes two or three, plus an annoyed customer. Improving first-contact resolution lowers your numerator and your denominator at once. Look at your slowest response times and your most-reopened topics; that is where the rework hides.
Fix the knowledge base, then point customers and AI at it. Most repeat contacts trace back to an answer that was missing, wrong, or buried. A help centre that is actually current gives any automation something accurate to stand on.
Automate the high-volume, low-variance topics. Order status, return policy, shipping windows, password resets. These are cheap to answer well and expensive to answer with a human's time. Routing them to automation is where cost per ticket genuinely drops. The topics full of exceptions are a different story.
Watch what automation actually costs. Cheaper-per-ticket is only real if the automation resolves the issue. If it produces an answer that triggers a follow-up, you paid for the automation and the human cleanup. We have written about this trap in the context of per-resolution pricing: a vendor's definition of a solved ticket is not always your customer's.
Reduce ticket volume at the source. The cheapest ticket is the one never created. A confusing checkout, a vague shipping page, a misleading product photo: each one manufactures tickets. Support cost per ticket and product quality are connected, even though they sit in different budgets.
How Keloa approaches this
We sell AI agents for customer service and sales over email and a chat widget, so we have a stake in this. We try to be straight about where the stake is.
Keloa lowers cost per ticket in two places. It resolves the high-volume, low-variance questions outright, so a human's loaded hour is spent on the work that needs judgement. And it grounds answers in your knowledge base rather than improvising, which keeps repeat contacts down, the lever that matters most.
On the cost side, we charge a fixed monthly plan with a bucket of replies and a flat top-up price after that. You can read the meter and forecast the bill, so the AI portion of your cost per ticket is a number you can plan around rather than a surprise. The full breakdown is on the pricing page.
We will also say the unglamorous part. Automation does not make every ticket cheaper. For tickets thick with exceptions, a human is still the lower-cost path once rework is counted. A tool that pretends otherwise is selling you a number, not a result.
Frequently asked questions
What is a good support cost per ticket? There is no universal target. It varies by industry, channel, and how complex your product is, and a higher cost per ticket can be entirely healthy if it buys higher quality. As MetricNet puts it, "a higher than average Cost per Ticket is not necessarily a bad thing, particularly if accompanied by higher than average quality levels." Track your own trend over time instead of chasing someone else's number.
Should I use tickets received or tickets resolved as the denominator? Resolved is the more honest choice, because an unresolved ticket still consumed cost and often returns as a repeat contact. Whichever you pick, define it once and never let it drift, or your month-to-month trend becomes meaningless.
Why is my cost per ticket higher than I expected? Almost always because the first estimate counted only front-line agent salaries. Add employer payroll taxes and benefits, the prorated time of leads and managers, every tool in the stack, and a share of overhead, and the number routinely lands well above the napkin figure. That is the real number, and it is the one worth managing.
Does channel affect cost per ticket? Yes. Phone support tends to cost more per ticket than chat or email because it is one-to-one and time-bound, and the spread between teams on the same channel is wide. The bigger lever, though, is not channel choice but first-contact resolution, since a repeat contact costs again regardless of the channel it arrives on.
Will AI automatically lower my support cost per ticket? Only for the right tickets. Automation lowers cost on high-volume, low-variance questions where it resolves the issue cleanly. On exception-heavy tickets it can raise total cost, because you pay for the automation and then for a human to fix the result. Measure cost per resolved ticket by topic before assuming automation is cheaper everywhere.
How often should I recalculate it? Monthly is enough for the trend, and the trend is what matters. Recalculate sooner after a real change: a new hire, a tooling switch, an automation rollout, or a spike in volume.
If you want to see where your own cost per ticket is leaking, start with the Free Starter plan, connect your help centre, and watch which topics automation genuinely makes cheaper.