The real cost of a ticket

Someone in finance highlights one cell in a spreadsheet and asks you to explain it: cost per contact, up again this quarter, answer wanted by Friday. So you divide total support spend by total contacts and hand back a tidy figure. Everyone nods. The number is wrong — or at least it hides almost everything that actually matters.

The metric everyone quotes and no one trusts

Cost per contact is the most cited and least understood number in support. The headline is easy: take what you spend, divide by how many tickets you close. But that clean quotient buries the two things that really move it — how you staff, and how long each contact takes — and it says nothing about the channel mix underneath.

The received wisdom is a tidy ladder: self-service is nearly free, email and chat cost a little more, phone sits at the top. It's directionally true and operationally useless, because the ladder assumes every contact on a channel costs the same. A chat that resolves in one exchange and a chat that limps across three agents over two days both count as "one chat" in the denominator. They are not the same expense.

What actually sets the per-contact cost isn't the channel — it's handle time and how many people touch the ticket. Which is where resolution speed stops being a customer-experience metric and becomes a line item.

That gap is the difference between a ticket that costs a few minutes of one person's attention and one that costs hours across a queue. Same channel, wildly different cost. The same research found early AI adopters cutting ticket volume outright — fewer contacts in the denominator, the one lever that improves the number without squeezing anyone.

The lever nobody writes down

Here's the part the spreadsheet never shows. When cost per contact has to come down, the fastest lever is occupancy — the share of paid time an agent spends actually handling contacts. Push it up and the math improves immediately, because you're spreading the same salary across more tickets.

That 83% isn't an accident; it's roughly the ceiling before the wheels come off. Above 90%, the staffing guidance is blunt: burnout becomes the most obvious consequence. And burnout carries a price tag that never lands in the cost-per-contact formula.

Turnover is the real hidden cost of a ticket. Every point of attrition is a recruiting cost, a training cost, a ramp period where new hires are slow and error-prone, and a tenure loss that quietly drags down first-contact resolution and CSAT. Half of contact-center leaders say their people left simply to chase better opportunities. So when you cut cost per contact by running agents hot, you aren't saving money — you're moving it off this quarter's spreadsheet and onto next year's hiring budget.

What the headline leaves out

Then there's the demand side the internal metric ignores entirely. Customers expect faster answers than they did a year ago, and a growing share expect service around the clock. Coverage isn't free. Every hour you staff to honour a 24/7 promise is capacity you pay for whether the queue is busy or not — and it sits in the numerator without ever showing up as a "contact."

This is why AI has become the story it has. It doesn't just deflect tickets; it changes what a contact costs to handle in the first place, which is why so many teams now say it's rewriting the economics of the whole function.

Whether that reshaping is as dramatic as the vendors claim is a separate argument. What isn't in dispute is the direction: the cost of a ticket is splitting into the part a machine can absorb and the part that still needs a person — and the person's part is the expensive, escalating, complicated one.

Costing a ticket honestly

So what do you tell finance? Not a single number. Give them a range and the assumptions behind it: fully loaded labour cost — salary plus the recruiting and training you'll spend to replace the people who leave — divided by contacts, and segmented by whether a ticket resolved on first touch or bounced. Track occupancy beside it, so nobody celebrates a falling cost per contact that's actually a burnout warning in disguise.

A ticket's real cost isn't what you paid an agent for the minutes they were on it. It's that, plus the fraction of a resignation you triggered, plus the ramp time of their replacement, plus the coverage you hold idle for the promise of a fast answer. The headline is a starting point. Everything interesting is in what it leaves out.

If you're benchmarking, compare like with like rather than relying on one blended average — the benchmark library shows how figures shift by channel and segment.