The first 90 days

TL;DR

The First 90 Days: Why New Call Centre Agents Leave So Soon

  • 12% of US centres lose over a quarter of new agents within three months; some lose 16% before day one.
  • A quarter of frontline hires say the job was not described accurately - and accurate briefing doubles retention likelihood.
  • Competence takes about six months; the danger zone is the gap between end of training and genuine confidence.
  • Replacing one agent costs $10,000-$20,000; early leavers cost that with nothing returned.

One operator hired 1,153 agents. Only 967 showed up for day one. That is 186 people - 16% of the intake - lost before a single call was taken, before onboarding, before anyone could ask why.

Early attrition is the most expensive kind. A leaver at month two has consumed recruitment, training and licence costs - McKinsey puts replacement at $10,000 to $20,000 per agent - and returned almost nothing. Yet the causes of the first-90-days exodus are unusually well documented, and unusually fixable.

Why do new call centre agents quit in the first 90 days?

Three causes dominate the evidence. First, the job was misdescribed: COPC research found 25% of frontline staff did not agree the recruiter accurately described the role, and accurately briefed hires reported twice the satisfaction and twice the likelihood of staying. Second, the gap between training and real competence - agents typically need six months to perform at the level of an average colleague, and the hardest weeks come right after training ends. Third, the hire was a poor fit for the role's real demands, which had gone unmapped.

The problem starts at the advert

The COPC finding is worth sitting with. One frontline worker in four says the job they got was not the job they were sold. Recruiters under volume pressure soften the picture - the shift patterns, the call types, the monitoring - and the gap between promise and floor becomes the new agent's first discovery. The same research shows the fix pays twice over: hires given an accurate picture report double the satisfaction and double the likelihood of staying.

The pre-start ghosting is part of the same story. A candidate who accepts and never appears has usually kept interviewing, and chose the offer that felt more truthful or arrived faster. Speed matters here, but so does the quality of what was promised.

The competence cliff

Training bays are safe. The floor is not. SQM's research puts the journey to average-agent performance at six months or more - which means an agent leaving training at week four faces perhaps twenty weeks of being measurably worse than everyone around them, on a dashboard everyone can see.

That stretch is where confidence either builds or breaks. Centres that grade call complexity - simple queues first, escalating as skill grows - and keep coaching close during the gap hold more people through it. Centres that graduate agents from training straight into the full queue, then manage the resulting numbers with performance conversations, convert their own onboarding into an exit programme.

What is a good early attrition rate for a call centre?

ContactBabel's 2024 US data gives the distribution: 45% of contact centres lose 5% or fewer of their new agents in the first three months, 27% lose 6-10%, 16% lose 11-25%, and 12% lose more than a quarter of new hires within 90 days. If more than one new agent in ten is gone inside three months, the recruitment and onboarding pipeline is leaking money - the same report puts median annual attrition at 31%.

The spread in those numbers is the interesting part. Nearly half of centres keep early attrition at 5% or below - so the sector's worst figures are not an iron law of the industry. They are the output of particular recruitment and onboarding choices, made by centres competing in the same labour markets as the ones holding people.

Fixing the pipeline

How do you reduce attrition in a call centre?

Start before day one: describe the job accurately, since accurately briefed hires are twice as likely to stay, and expect some pre-start loss - one operator recorded 16% of accepted hires not appearing for day one. Then shorten the distance from training bay to competence with graded call complexity and close coaching. Longest lever: hire for fit against what proven performers in the role actually look like, rather than against a generic profile, and read new joiners' preferences early so coaching lands the way each person needs it.

The last lever is the least practised, because most centres have never seen what their proven performers have in common. Job specifications describe an idealised agent; the people who actually stay and perform often look quite different. Where a role's successful holders cluster - in how they prefer to connect, decide and deliver - that cluster is readable, and it changes who you hire and how you onboard them.

This is what a successMAP does in Sariio: it reads the working preferences of a role's proven performers from the same ten-minute survey everyone takes, and shows whether they cluster. When they do, recruitment gains a benchmark grounded in evidence rather than adjectives - and onboarding gains something just as useful, because a new joiner's own reading shows their team leader, from week one, how this particular person prefers to learn, take feedback and build confidence. The survey is retaken at least twice a year, so the picture moves as the person settles. No verdicts, no fit scores - a shared reading, at the moment it helps most. The wider case is in why call centre agents really quit.

See what a new agent's first reading looks like - the individual survey is free.

Sources