The Contact Center Attrition Test

“They just use your mind and they never give you credit/It's enough to drive you crazy if you let it” —Dolly Parton (“9 to 5”)

Companies love to tout the benefits of their artificial intelligence spending and deployments. It’s like they are taking a prideful gym selfie after a single push-up. Their investment may be real, but the flex usually arrives well before the transformation. To make matters worse, or at least more complicated, in tech today the results being touted are often headcount reductions. That makes the victory lap feel a little…ummm…icky. It seems less like a celebration of better work and more like a tour of the empty desks left behind.

“How many people did we fire” is, thankfully, not the only way brands measure success in AI. In the customer experience world, we often see companies crowing about lowered service costs through reduced average handle time; reduced cost per contact; fewer repeat contacts; and more self-service resolution. (Speaking of icky, can we stop calling this containment or deflection? CX is supposed to start from the customers and their needs; these terms are fossilized versions of inside-out thinking.) We also see companies citing revenue increases from improved cross-selling, upselling, conversions, recovery of abandoned journeys, and higher shares of wallet. And in the regulated world, we see companies point to risk reduction, including fewer compliance failures, fewer complaints, better audit trails, and overall fraud reduction.

Those are all decent measures. But decent measures aren’t the same as a complete proof-of-value story, especially when the loudest metrics make the organization look efficient before they prove the customer is better served. And those metrics often overlook the very employees that are supposed to be delivering the high-touch, high-EQ work left behind because AI couldn’t handle it. So here’s one measure I’d like to see get more attention: contact center attrition rates going down. Not because AI made the job easier in the lazy sense but because AI made the job better.

For decades, the agent role has been designed around endurance. Take the next call. Find the right answer. Calm the customer down. Document everything. Stay compliant. Keep the queue moving. Be measured within a micrometer of your life. Repeat until your headset melts into your skull. Ain’t no wonder the job has absurdly high churn baked into it.

But if all these tech vendor promises are true, AI changes the center of gravity. If AI can handle search, summarization, authentication support, after-call work, routine troubleshooting, and repetitive case types, the human role can move up the value chain. Agents become problem solvers. They can spend more time interpreting context, weighing tradeoffs, negotiating exceptions, and restoring trust when the process has already failed.

But to do that well requires greater skills. The work becomes less about reading from a script and more about knowing when the script doesn’t fit. It asks agents to understand products, policies, systems, customer history, and business priorities. AI can tee up the facts, but the person still must decide what matters.

It also requires higher EQ. Once the easy stuff is automated, the remaining interactions are often more emotional, ambiguous, or high stakes. Customers don’t want a warmer transfer to a robot. They want someone who can hear frustration, explain choices, and make a judgment call without sounding like they’re being held hostage by the policy manual.

And in some industries, this opens the door to more credentialed roles. Healthcare, financial services, insurance, travel, benefits, education, and government services all have moments where customers need more than generic support. They need licensed, trained, or deeply specialized people who can operate with AI as a force multiplier instead of a compliance risk.

That’s why attrition is such a useful test. If AI is really improving CX work, people would stay, grow, and build careers around it. Lower attrition would mean AI didn’t just make service cheaper. It made the job more sustainable, more skilled, and more worth keeping. Now that’s a proof point worth flexing.

Ian Jacobs is vice president and lead analyst at Opus Research.

Related Articles

Bots Get Agency, You Get Monitored

The machine gets to be the agent. The person gets to be the backup plan.

Cold Circuits, Warm Heart: AI and Empathy in CX

Customers will judge the interaction in front of them.

Profiled by Prompt, Illustrated by Model: Audit Your LLM’s Assumptions

Your large language model thinks it knows you. Is it right?