I’ve been in the room when it happens.
The quarterly CX report is strong — service level is excellent, retention is up and satisfaction scores are holding. And then someone, usually the person financially accountable to leadership, asks out loud: what did that cost us?
Usually that question isn’t a big deal. Contact center and CX operators have a counterpart in finance, sometimes the CFO, who handles that answer. And that finance relationship has let contact centers focus on the activity that took place and the projected benefits the activity generated.
I can’t recall a single contact center where this relationship wasn’t good. But that dynamic is poised to change as AI becomes more deeply embedded into contact center operations.
AI is the Catalyst, Not the Cause
What’s changing for contact center and CX leaders, if it hasn’t already, is taking on cost accountability. What I mean is being responsible for generating the cost answers finance and others used to handle.
This sounds like long-awaited good news, that your CX operation is finally being taken seriously as a value driver. But the answers you’ll reach for won’t change. It’ll be an average cost — one number with a dozen different things folded into it. A 30-second touch and a 12-minute escalation, for example, can end up priced the same in an average cost model.
That’s what AI is revealing — contact centers and CX have become exceedingly complex operations: a single contact involves a multi-layered approach as it moves across self-service, a bot, an IVR path, a chat queue and a human specialist, each running at a different cost with different durations by engagement.
It’s in this complexity that leadership needs to know the actual cost of work before AI was deployed, at the precise level and location the AI is meant to work. Seen in this way, the reason cost accountability is moving to you is because finance can’t measure actual costs, and you are the leader nearest to the deployment.
This change isn’t anyone’s fault. This is business evolution and it’s worth being viewed through a new lens now that AI has entered the building.
Average Costs Were Built for a Simpler Operation
Average costs worked well when contact centers were mostly one channel, with similar contacts handled by agents at similar cost. And while these averages have served for longer than anyone’s career, the introduction of AI has made something clear.
Puzzel’s State of Contact Centers 2026 survey across the UK and northern Europe, shows how complex things have become: the average operation runs 3.9 different technology solutions, with only 3% on a single unified platform. And each one carries its own unique cost.
I saw that level of complexity hiding inside a contact center that was part of a subsidiary in a large enterprise. Leadership was under a hard profitability mandate, and the contact center was asked to commit to budget reductions.
The team had already worked with finance on time studies and average cost per contact. But the contact center was handling multi-channel work, including off-phone activity, and the CIO and COO suspected there were savings buried. That’s when they called me.
The stakes were high. The business unit CEO needed hard evidence for the board, and all contact center projects were paused until they had it. This wasn’t a punishment, but a lifeline — either the contact center could find savings that would withstand C-suite scrutiny, or leadership would decide the budget cuts for them.
I rebuilt the costs from the ground up, queue by queue. No averages. Measured in this way, $3 million dollars in savings came into view — savings that operational complexity had hidden inside average cost models. That was on top of $4 million already committed, with no decline in retained revenue or customer satisfaction.
The money wasn’t new, and nobody was trying to hide it. It was invisible because operational complexity had exceeded what average costs were designed to measure.
That’s the gap that AI is exposing. It shows up as real savings sitting unseen while the pressure to find them moves toward you — until someone is one board meeting away from having decisions made for them.
Closing the Gap Between Activity and Cost
Any real cost answer requires you to first recognize how cost answers are generated today in relation to the activity metrics you rely on to run your business.
Activity is measured from the ground up starting with the interaction and the steps inside it, at the location the work happens. Cost has been measured from the opposite direction — a top-down total cost divided into an average. And while we instinctively treat these metrics as equivalent, they don’t meet at a point of equivalence. Activity is actual and precise, while cost is an estimated allocation.
A real financial answer closes that gap. It builds cost in the same direction the activity is already measured: from the ground up, not from a total carved into smaller pieces.
A Smaller Average is Still an Average
The natural reflex, when an average stops being precise enough, is to slice it finer — cost per minute instead of cost per contact, cost per second instead of cost per minute. But a smaller average is still a slice of the same pie.
And it’s the starting point that decides the result. Begin with a total and divide, and you get an allocation no matter how fine the slice. Begin with the actual cost of the specific resource doing the actual work, and you get something an average can’t give you: a number that answers the question being asked at the level it’s being asked.
AI has shown that the level at which questions are asked has moved. AI doesn’t optimize a contact center in general. It optimizes a specific task, in a specific place, inside a specific interaction — layers below what an average cost can see. Again, it’s not what an average was designed to do.
Measure the Work, Not the Average
Here’s the part worth remembering: this isn’t a call to collect more data.
Customer activity data has always originated at the interaction — that’s where it comes from, and where it always has. What was missing was never the data. It was the cost, built to the same level. The raw material is already sitting in the systems you pay for today; it has simply never been constructed from the ground up to match the way the activity is measured.
That’s what a real financial answer requires. Not a new metric to chase or a dashboard to install, but a rebuilding of the cost side of the equation so it matches the level of resolution the activity side already has.
So, the next time someone in that room asks what work actually costs, there’s a version of the answer that doesn’t begin with “on average.” It begins with the work — and the actual work has been measured all along.