And it’s not because you haven’t tried everything…
If you lead CX, you’ve almost certainly asked your provider for lower attrition. You’ve tightened the SLAs, funded better training and asked for a retention plan.
All the right instincts. They’re also not enough on their own, because the quality problem keeps coming back.
The effects tend to show up elsewhere first. CSAT may slide after a resignation wave, or a newer agent may know the process without quite sounding like your brand. Meanwhile, the attrition line in the QBR can look fine even as customer transcripts point to a quality problem.
The cost goes beyond the empty seat
The recruiting cost to refill the seat is the obvious part. Training, ramp time and the supervisor hours pulled into coaching add to it. Then there’s the harder cost to quantify: the customer who reaches a new agent and gets an answer that’s almost right.
One contact-center analysis put the cost of attrition between $10,882 and $22,691 per agent, depending on when the employee leaves, after accounting for recruiting, training, nesting and lost productivity during replacement ramp-up.
Why a lower rate isn’t enough
Attrition is part of running a contact center. A 2025 survey from NICE and Simpler Media Group found average agent attrition was 39% in 2024. In EMEA, it was 41%. The survey included 400 contact center leaders, split evenly between North America and EMEA.
Even the strongest, best-run contact centers lose people, and in some markets they lose them for reasons outside what a coaching plan can achieve.
In EMEA, a single program might run in six languages. When a Finnish-speaking agent working from Portugal resigns, the replacement isn’t waiting in the local talent pool. Backfilling takes time and money. Simply lowering a target doesn’t change that.
People will leave, so what matters more is how the operation responds before, during and after that turnover.
What to measure instead
Attrition rate used to tell you something useful. In a model where every replacement meant a long training cycle and a slow ramp, a high rate really did mean a shaky operation. Watching attrition was the closest thing to measuring stability.
That link is weakening. When onboarding is compressed and attrition risk is caught early, a program can manage turnover and still hold the line on quality. The rate tells you less about the outcome.
Two numbers tell you more:
1. Time to proficiency. How quickly a new agent reaches full productivity. Not how fast they finish training. How fast it takes them to get really, really good.
2. Performance consistency. Whether the standard holds across the whole workforce, regardless of tenure, site, or season. At Foundever, this is managed by measuring performance variance between the strongest and weakest performers on a team. Tight variance means a customer gets the same experience whether they reach a two-week agent or a two-year one. Attrition widens that spread, so the challenge is to close it fast.
What changes when the system expects it
Picture the same resignation two ways:
The old way
Think about the traditional lifecycle of an empty seat: an agent quits, and performance immediately drops while the chair sits empty. Quality scores decrease, and predictably, the client flags it in the next QBR.
We have certainly gotten significantly smarter about anticipating this. By staying ahead of the curve, predicting attrition patterns, mapping out seasonal demand, and keeping a close watch on local labor markets and competitor moves, we can drastically cut the time it takes to source and hire a replacement. That kind of operational intelligence is important, but filling a seat fast doesn’t flatten the learning curve.
You still have to survive a full classroom cycle, followed by a slow, highly variable ramp phase where new agents are live on the phones but not yet performing at full strength. Even with the sharpest, most proactive recruitment engine behind you, by the time performance recovers, the impact has likely already shown up in the experience of your customers. Faster hiring solves the vacancy problem, but it doesn’t solve the proficiency problem.
A better way
Culture and role fit are screened up front with a Foundever assessment methodology, so a poor match can be caught before it turns into an early exit. Onboarding uses AI simulation to let agents practice difficult calls, like billing disputes and high-stress moments, before they are in a live call environment.
Then there’s STAR, or Safe Talent At Risk. It’s a defined, tracked process built into monthly coaching. Team managers score each agent’s risk, investigate what’s behind it and, when someone is flagged, build a save plan that can involve learning, HR, workforce or other support.
That focus on faster proficiency already has measurable results.
A fintech program in the German market rebuilt its new-hire path around getting agents productive sooner. Time to competence dropped 25%, from 16 to 12 working days. In their first week on the phones, agents trained under the new approach averaged a 56% NPS compared with 29% for agents trained using the traditional method.
When the customer can’t tell the difference
The goal is an operation where a customer can’t tell a new agent from a seasoned one because the new agent gets good fast enough that the experience doesn’t slip.
Building that kind of flexibility starts long before someone leaves.
Attrition will always be part of the picture. The question is whether your operation is built to absorb it without the customer ever noticing. If you want to see how CX leaders across eight industries are approaching that challenge, “The CX complexity guide” is a good place to start. It covers what the furthest-ahead leaders are doing differently on AI, cost, and agent retention, plus the questions you should ask any partner before you sign.