Call Center Turnover Rate: Benchmarks and What Fixes It

Every page ranking for this term is published by a company selling contact center software, and they all give you the same kind of number and the same list of causes. A turnover range with no survey named behind it, then pay, then burnout, then career development, then a call to action.

Here is something none of them mention. The US Department of Labor publishes a fact sheet dedicated specifically to call centers, and its section on typical problems reads like a list of reasons agents quit. It states that starting the computer to download work instructions and applications is the first principal activity of the day for call center agents, which means it is paid time. It states that short rest breaks must be counted as hours worked. It notes that salaried call center employees sometimes do not meet the tests for being exempt. Every one of those is simultaneously a wage and hour exposure and a thing agents talk about in the break room before they leave.

This guide gives you the benchmark and the formula, then explains why published figures disagree so widely without any of them being wrong, then covers what federal data actually says about this workforce, the pay rules written for it, what a departure costs a small team, and the six drivers you can change. I build the time records, onboarding, and employee files this runs on at FirstHR. This is general information rather than legal advice, and state rules add to the federal baseline.

What the Call Center Turnover Rate Is

The call center turnover rate is the percentage of agents who leave a contact center operation over a given period, measured against the average number of agents during that period. It is used to benchmark an operation against the industry and to size the cost of replacing people.

One structural feature makes this metric behave differently from turnover in most other roles. Some agent departures happen before the person is productive at all, during classroom training or the supervised handling period that follows it. Those departures cost the full training investment and return nothing, and they are usually invisible in a headline turnover figure that treats every leaver the same. Counting them separately is the first upgrade most operations can make to their reporting.

How to Calculate It

The arithmetic is simple. The decisions around it are where the confusion lives.

For a team under thirty agents, the monthly view matters more than the annual one. At twelve agents a single departure is an eight-point move, so an annual figure smooths away exactly the timing information you need.

The Average Call Center Turnover Rate

The best-documented benchmark is typical turnover of 20 percent a year, from the Global Call Center Report, a Cornell-led survey of almost 2,500 centers in 17 countries. In liberal-market economies such as the US, the median was 25 percent.

For the wider economy, the Bureau of Labor Statistics measured a monthly quits rate of 2.0 percent and a total separations rate of 3.3 percent across 2025, roughly 24 percent of employment in voluntary turnover and 40 percent in total separations over a year. Industry articles quote call center figures well above the research benchmark, usually without naming a survey, a sample, or a date.

The Cornell figures were published in 2007, so read them as structure rather than as a current level: what separates low-turnover centers from high-turnover ones has not changed, even if the labor market around them has.

Two corrections to what you will read elsewhere. First, the direction of travel: annual quits across the US economy reached 50.6 million in 2022, per a Monthly Labor Review analysis, and fell to 38.0 million by 2025. Pages telling you turnover is rising are often quoting numbers from that peak.

Second, the in-house gap. In the Cornell data, in-house centers turned over fewer people than subcontracted ones, so a headline drawn from outsourced operations overstates what a small in-house support team should expect. Beating that kind of headline is a low bar for an in-house team of fifteen.

Why Published Call Center Turnover Figures Disagree

The spread in published numbers is not a sign that somebody is wrong. It is five definitional choices, each of which moves the answer, and almost no page states which ones it made.

The practical consequence is that comparing your number against a figure you found online is nearly meaningless unless you check the definitions first. A research median of total turnover, a figure an outsourced operator publishes, and a voluntary-only figure can all describe operations similar to yours while landing far apart.

What Federal Data Actually Says About This Workforce

Industry surveys dominate this topic and federal data is barely used, which is a shame, because the federal picture reframes the whole problem.

Per the Bureau of Labor Statistics occupational outlook for customer service representatives, employment in the occupation is projected to decline 5 percent from 2025 to 2035, and the median hourly wage was $21.53 in May 2025. And yet the same projection expects about 289,500 openings each year, on average, across that decade.

All of those openings are expected to come from the need to replace workers who transfer to other occupations or leave the labor force, not from growth.

Two further federal data points give useful context. The employee tenure survey shows about 22 percent of all US wage and salary workers had a year or less with their current employer, with median tenure among workers aged 25 to 34 at roughly 2.7 years, as of January 2024. The Cornell survey found 21 percent of call center workers in liberal-market economies had less than a year of tenure, close to that all-worker share, although the two were measured almost two decades apart.

The job openings and labor turnover survey reports separations as a monthly rate, which is the source of the monthly-versus-annual confusion described above.

The projections commentary is worth reading for what it says about the cause. The handbook attributes the decline to tasks continuing to be automated, as self-service systems, social media, and mobile apps let customers handle simple requests without a representative. The 2025 to 2035 projections release adds that automation tools, including those powered by AI, are likely to reduce demand for several office and administrative support occupations.

What that means for a manager is that the job is changing under the people doing it, and roles that survive automation are the harder ones, which raises the demands on the agents you keep while the headcount around them shrinks.

The Pay Rules Written Specifically for Call Centers

This is the section missing from every page ranking for this term, and it is the only one on this page with legal exposure attached as well as turnover cost.

The Department of Labor publishes a fact sheet devoted to call centers, and its typical problems section is effectively a list of the ways this industry gets paid time wrong. Per Fact Sheet 64, an example of the first principal activity of the day for agents working in call centers is starting the computer to download work instructions, computer applications, and work-related emails. That places the boot-up sequence inside the paid day rather than before it.

Whether any of that is happening in your operation is not a question you can answer from memory. It has to be timed, on real agents on real mornings, and compared against the window your timekeeping system actually pays for.

The audit above takes an afternoon and produces two things at once: a number for how much unpaid time your operation is running on, and a fix list. The most common single fix is moving the clock-in to the moment the agent powers on the machine rather than the moment they reach the timekeeping screen.

Classification deserves a separate note because the fact sheet calls it out for this industry specifically. A team lead paid a salary who spends most of the day handling contacts alongside everyone else may well be not exempt, whatever the title says. The consequence of getting it wrong is unpaid overtime owed for up to two years, or three for a willful violation, plus an equal amount in liquidated damages.

Why Call Center Turnover Runs So High

The published cause lists are all the same and all incomplete. Here is the version that includes the item nobody writes down.

The metrics point deserves expanding because it is structural rather than cultural. An agent scored on average handle time and on customer satisfaction has been handed two objectives that pull in opposite directions, and the honest response to a confused customer is the one that damages their scorecard. Operations that pick one primary measure and treat the other as a guardrail see the tension disappear. Operations that weight them equally have built a job in which nobody can fully succeed, and being unable to succeed is a better predictor of resignation than pay is.

What Agent Turnover Costs

The best-documented estimate puts the simple replacement of one agent at about two months of pay, and three to four months once lost productivity is counted. That comes from the Global Call Center Report, and the biggest item in it never appears on an invoice.

That item is the ramp. The same report put the time for a new agent to become proficient at 16 weeks in liberal-market economies such as the US, during which they are paid in full, consume trainer and supervisor time, resolve fewer contacts, and escalate more of them.

For context on the wider hiring side, SHRM benchmarking data for 2026 puts the median cost per hire for nonexecutive roles at $1,300, and that is before anything about the departure or the ramp is counted.

The way to get your own figure is to record the departures as they happen and price the pieces from your own payroll, rather than adopting somebody else's average. Three sheets cover it.

The first sheet gives you monthly and annualized rates from one set of inputs, which removes the comparison problem at source. The second builds cost per departure bottom-up, with a separate line for the supervised handling period that most calculations omit entirely. The third is the exit log, and its most important column is days employed, because a team losing people at week three has a completely different problem from one losing people at month fourteen.

If You Run a Support Team of Five to Thirty People

Nothing in the top results for this topic is written for you. The published advice assumes a workforce management department, a quality assurance function, and a cost table denominated in millions.

The ramp row is the one small teams underestimate. In a large operation a new agent is trained by someone whose job is training. In a fifteen-person team the trainer is your best agent, so every hire costs you the output of your strongest person for several weeks on top of everything else. That makes the case for reducing turnover stronger for you than for a large operation, not weaker.

The advantage to press is the last-but-one row. You can ask. A five-minute conversation with every agent at day 7 and day 30 will tell you more about your retention problem than any published benchmark.

Remote and Hybrid Agents

Home-based agent work became normal during the early 2020s and stayed, and it changes the turnover picture in both directions.

The monitoring row is the one that quietly costs the most. Call recording is expected and agents accept it. Screen capture, keystroke logging, and camera checks are a different category, and the gap between what quality assurance actually requires and what the software happens to offer is where resentment accumulates. Our guide to employee monitoring laws covers what is permitted, which is a different question from what is wise.

Onboarding remote agents is the other place small teams lose people. Somebody who spends their first week alone on a video call with no one to ask a quick question of has a materially worse experience than the same person in a room, and the fix is structural rather than motivational.

What Actually Reduces Agent Turnover

Ordered by return per dollar, weighted toward what a small team can do without a budget.

Six of those eight cost approximately nothing. That matters because most exits can be headed off: the Work Institute 9th Annual Retention Report classed 75 percent of 2025 departures as preventable, drawing on more than 120,000 exit interviews.

Training deserves a closing note because the arithmetic is unusually favorable. An agent who leaves in week three consumed the entire classroom investment and returned nothing, so anything that improves completion is pure recovery. A written onboarding checklist and structured onboarding training take a day to set up and then run themselves.

What to Track

Five numbers, reviewed monthly, cover any operation under fifty agents.

Start with the second row. Losing agents before they ever handle a contact unaided is the most expensive kind of turnover per head and the least visible, because those people rarely appear in anyone's mental model of the team. If a fifth of your departures happen before the ramp completes, the problem is upstream of retention entirely.

Where Operations Get This Wrong

The same errors, across in-house teams and large operations alike.

Comparing against a benchmark without checking its definitions is first. A research median, a figure an outsourced operator publishes, and a voluntary-only figure can all be real, and none of them is your number.

Not paying for time around the shift is second, and it is the one with legal exposure. Federal guidance names boot-up as the first principal activity of the day for call center agents, and short rest breaks as hours worked.

Scoring agents on conflicting metrics is third. Handle time and satisfaction pull against each other, and a job in which nobody can fully succeed produces resignations regardless of pay.

Treating departures during training as noise is fourth. They cost the full investment, return nothing, and are the most preventable category available.

Setting occupancy for maximum efficiency is fifth. The recovery seconds you removed cost less than the turnover they produce, and the trade is rarely calculated.

Deploying every monitoring feature the software offers is sixth. Call recording is expected; keystroke logging and camera checks are a different thing and they cost you people quietly.

Assuming a salary makes a team lead exempt is seventh. The federal call center fact sheet lists it among typical problems, and duties decide it rather than the pay method, with retrospective payroll records being what an audit looks at.

And treating high turnover as the nature of the work is last. It is common in this industry, which is a statement about the industry rather than a verdict on your operation. In the Cornell data, centers with high-discretion, lightly monitored jobs ran typical turnover of 9 percent against 36 percent where work was tightly scripted and heavily monitored, which says job design moves the number more than the industry label does. At larger scale the scheduling side gets handed to call center scheduling software; below that, a forecast, a spreadsheet, and the discipline to run the exit log weekly do the same work.

Frequently Asked Questions