A support agent who starts in January is doing the job at full speed by August and, on the industry's own averages, has left by March of the next year. That is not a bad hire. That is the average one. The two figures behind that sentence come from the same publisher, and together they say something about hiring that most turnover advice skips: by the time a new agent is fully productive, half of the stay is already over.
This article reads the support agent turnover numbers from the hiring side. Not how to keep agents once they are in the role, which is a management job with its own article, but what the numbers say about the decisions made before the offer, and which of those decisions actually move how long a person stays.
Support agent turnover in numbers: half of the stay is below full speed
Start with the two numbers. Insignia Resources is a staffing firm that supplies support teams from abroad. Its 2026 call-centre benchmark puts the time a new agent needs to reach the performance level of experienced staff at 6 to 8 months. That period is the ramp, and it covers everything the job takes to do well: the product, the tools, the queue, the escalations, the judgement calls. Its companion page on customer service roles puts the average stay for customer support specifically at 13.7 months. Put those side by side and the arithmetic is uncomfortable: of a stay just under fourteen months, six to eight are the ramp. On the average, the time at full speed is about as long as the time it took to get there, and then the role is open again. For many agents it is shorter than that: most of the leaving, as the figures below show, happens inside the first year, so an average of 13.7 months is raised by the people who stay for years.
The scale of the leaving is on the same page. Insignia Resources puts call centre turnover at 40-45% a year in 2026: over a year, a team of a hundred people records forty to forty-five departures. First-year attrition, the share of new hires who leave before their first anniversary, sits at 69-73%, which the report reads as most turnover happening inside the first twelve months. The two figures count different people: the first is the whole team across a year, the second is that year's new hires. The picture is not a slow drain across a career. It is most of the departures before the first anniversary.
Those figures come from a firm that sells to American companies, and the page does not say which countries they cover. The one European figure we found is a different kind of number: Call Centre Helper, citing ContactBabel's benchmarking of contact centres in the United Kingdom, reports that one in five of them has an attrition rate of over 30%. It does not compare directly with the averages above, and it is the closest thing to a European benchmark we found. A ten-person support team inside a software or fintech company is not a contact centre and has no benchmark of its own. We use these figures as the nearest measured proxy, not as a description of your team. What carries over is the shape of the problem: a ramp that has to be paid before the person is productive, and departures that cluster early. Shortening the ramp is a training question and belongs to the management article linked above. This page is about the other half of the arithmetic: how long the person stays, and which decisions made before the hire move that number.
The two things agents complain about are set on the employer's side of the table
The obvious response to those numbers is a retention programme: better coaching, a recognition scheme, a new quality tool. Look at what agents actually name when asked, and the response lands late.
Verint surveyed 1,000 contact centre agents at companies with at least 300 agents for its State of Agent Experience 2026 report, released in April. Nearly one third, 31%, said they were likely to leave their current role within the next six months. No Jitter, reporting the same survey, listed the two biggest frustrations across all respondents: unrealistic performance expectations, named by 47%, and lack of schedule flexibility, named by 45%. On the other side of the same survey, nine out of ten agents said schedule flexibility is important when choosing a job.
Read those figures together. The targets and the shift pattern are the two things agents complain about most, in a survey where one in three is already planning to leave, and the schedule is the thing nine in ten of them weigh before they say yes. Both exist before the agent arrives: the target system and the shift pattern are set on the employer's side of the table, and the hiring stage is where the candidate either learns them or does not: the ad, the first conversation, the offer. Neither is something the agent chose, and the hiring stage is the first moment at which a candidate can see them and the cheapest moment to say no.
The third figure fits the same pattern. Insignia Resources reports that only 22% of agents see a clear path for advancement. Fewer than one in four can see the next step from the role they are in, and the next step is another thing that exists, or does not, long before the hire is made.
Those surveys describe large centres, and a team of eight in a Berlin fintech is not one. We read the mechanism as the same, not the percentages: the two things agents complain about most were decided before they arrived, and the cheapest moment to get them right is while the role is still empty.
Rejecting candidates for short stays does not work
One filter against turnover sits at the hire, and it is the one the evidence is clearest about: a work history with several short stays goes to the reject pile, on the theory that people who left before will leave again. That theory has been tested at scale, and it failed.
In 2012 Forbes published an analysis by Evolv, a workforce analytics company, of over 20,000 hourly-paid employees. Its finding for a hiring team is direct: screening applicants on whether they had held several jobs for less than six months had no correlation with how long they stayed or how well they performed. The data is from 2012, the article does not say where the sample was drawn, and nothing newer we found overturns it. The screen looks like diligence. It removes candidates and adds nothing.
What the same analysis found did matter was a person's connection to the company: employees who knew three or more people already working there were more likely to stay. That is not a filter you can run on a document. It is a reason to spend the sourcing effort on referrals and on real conversations, which is where our post on screening candidates arrived for a different reason: when a resume can be written by software, the written application stops telling you much about the person.
An honest job preview helps a little, and costs one conversation
The lever with actual evidence behind it is old, cheap and undersold. A realistic job preview tells a candidate the bad parts of the job along with the good ones, before they accept. The Quality Improvement Center for Workforce Development, a research body at the University of Nebraska-Lincoln, summarised four meta-analyses of the published studies and is careful about the size of the effect: realistic job previews have only a very modest effect on job retention and job performance. Two details inside that summary are the useful ones. Previews given orally, face to face, reduce turnover more than written or video previews. And previews given after the hire work slightly better than those given before it.
So the honest version of this lever is not a paragraph in the job ad. It is a person, in the first conversation and again in the first week, saying what the queue looks like on a bad Monday, what the working hours actually are and which of them can change, what the targets are for month one, month three and month eight, and what the next step after this role is, if there is one. A few more people stay. Not many. It costs a conversation, so it still pays. In our reading its second effect matters more: the candidate it discourages withdraws before day one, which is the cheapest departure there is. The preview is not the argument of this article. The targets and the shift pattern are, and unlike the preview they are settled before the search opens; that is why the preview is the small lever here and the hiring brief is the big one.
Pay belongs in that conversation too, said as a number rather than a promise. The pay range does most of its work in the first exchange, and the reasons for that sit in our post on the offer acceptance rate: by the time the offer arrives, the decision has already been made.
Eastern Europe shows the lowest turnover in the benchmark
One more line from the Insignia Resources benchmark deserves more attention than it gets. Comparing the regions where support teams are staffed, the report finds that Eastern Europe shows the lowest turnover, 27-36%, even with mid-range compensation levels. Twenty-seven to thirty-six departures per hundred people a year, against forty to forty-five for the industry as a whole, and not because the region pays the most.
The report puts it down to cultural factors, strong education systems and professional work environments, but shows no data behind that, so we do not build on it. What we can say is where we hire from: the Ukrainian and wider Eastern-European market, and internationally. The region's own turnover number is a regional average, not a measurement of any one country, and not a promise about any one hire. A candidate from a lower-turnover market who takes a role with an unrealistic target and a rigid schedule leaves for the same reasons as anyone else. The region is a background rate. The role design decides the rest.
The guarantee ends while the agent is still learning
If a recruiter placed the agent, the contract usually has a clause for early departure: the guarantee, under which the agency replaces the hire, or refunds part of the fee, if the person leaves inside a set period. Here it is measured against the ramp. Read the dates on it against the numbers above. Top Echelon, surveying its own network of recruiters, found the most common guarantee period is 90 days, chosen by 44.9% of them. Ninety days is month three. The ramp runs to month six or eight. Most of the leaving, on the first-year attrition figure, happens across the twelve months. The guarantee closes while the person is still learning, and from month four to month twelve, the rest of the year in which most of the leaving happens, nobody is liable for a replacement.
This is why the replacement clause is a hiring question and not a finance one. What a month-five departure costs the recruiter is decided by the length of that clause: one that ends in month three costs them nothing, one that runs to the end of the ramp makes the second search theirs. A recruiter who carries the cost of a month-five departure has one more reason to shortlist people who will still be there in month six; one whose liability ends in month three has one reason fewer. So ask what a month-five departure costs them, and read the answer as one signal among several.
Five things to settle in the hiring brief before the search starts
None of this needs a retention programme. It needs five sentences in the hiring brief, the instructions handed to whoever runs the search, whether that is a recruiter or your own team, written before the first candidate is contacted.
- Why the last person left, in one honest sentence. If the answer is the schedule or the target, the role has a problem the next hire will inherit. Fix the role or say it out loud in the preview.
- The working hours, and how much of them is fixed. Nine in ten agents weigh this before they accept. Put the real pattern in the ad, including the weekend and the evening, and say which parts are negotiable. In much of the European Union the pattern is not the employer's alone: a works council or a collective agreement may own part of it, which is one more reason to have it settled before the ad goes out, not after the offer.
- The targets for month one, month three and month eight. In our reading, a single number for an experienced agent, applied from week two, is what makes a target unrealistic. Write the ramp into the targets.
- The pay, against the people already doing the job. The range will also be read by the team the new hire will sit next to, and they will compare it with their own pay. Price it against them before it goes out.
- What the next step is, if there is one. If there is a next step, name it in the first conversation. If there is not, hire someone for whom this role is the destination, and say that too.
And one thing to remove from the brief: the short-stay filter. Read what the person did in each job. Do not count the jobs.
How IMMIDO fits
We run a first-line support team ourselves: seven agents for Spribe, for more than five years, staffed by people we hired. We do not publish a retention figure. What we can say is that the pod has run for that whole time, and that seat is the one this article is about.
We place support and operations agents. The price is a flat fee, paid when someone is hired, and we deliver a first qualified shortlist in 7-10 business days. On recruiting, that is the only fee: paid at the hire, and nowhere else. Ask what our replacement terms are before you sign, the way you would ask any recruiter, and put the answer in the contract. Bring the five answers above to the brief, or let us ask for them there; a shortlist is only as good as the role it is built for.
If you would rather not run the team at all, staffing and running it as a managed team is a separate service, a fixed monthly team, not per ticket, and we will say which one you are buying.
If you are hiring a support agent this quarter and the last one left inside a year, see how our recruiting works or book a call and get a quote.
Sources: Insignia Resources, Call Center Turnover Rates: 2026 Industry Average and Customer Service Turnover Rate (22 April 2026); Verint, State of Agent Experience 2026 press release (14 April 2026); No Jitter, Agent attrition rates are high, Verint survey finds (15 April 2026); Forbes, Big Data Debunks View That Job-Hoppers Make Bad Hires (4 November 2012); Quality Improvement Center for Workforce Development, Realistic Job Previews (19 February 2020); Call Centre Helper citing ContactBabel (8 September 2023); Top Echelon recruiter survey (18 March 2019).