Attrition is 14% and that tells you almost nothing

A leadership team spent forty minutes on an attrition figure of fourteen per cent. Somebody said it was broadly in line with the industry. Somebody else said it felt higher than that. The conversation ended with an agreement to look at salary benchmarking, and nothing changed for a year.

When the number was eventually cut by team, two of eleven teams accounted for well over half the leavers. One had a new manager. The other had been running short-staffed for eight months, which meant everyone remaining was covering, which meant more of them left. Neither had anything to do with salary, and neither was visible in the headline.

Why the single number fails

Attrition is an average across populations with nothing in common. A twenty-two-year-old in their first job, a supervisor with eleven years' service, and a senior hire who joined four months ago are leaving for entirely different reasons, and one figure combines them into a number that describes none of them.

It also invites benchmark comparison, which is the least useful thing you can do with it. Knowing that your figure resembles a sector average tells you nothing about whether the people you could least afford to lose are the ones walking out.

The four cuts that matter

BY TEAM
  Almost always the first finding. Attrition concentrates.
  Two or three teams usually carry most of it, and the
  common factor is frequently one manager.

BY TENURE
  Under 6 months  -> hiring or onboarding failure
  6-18 months     -> role reality vs role promised
  2-4 years       -> no visible progression
  5 years+        -> life change, or long-held grievance

BY PERFORMANCE RATING
  If your strong performers leave at a higher rate than
  your weak ones, you do not have an attrition problem.
  You have a selection problem, and it is urgent.

BY REGRETTED / NOT REGRETTED
  Decided by the manager at the time of exit, not
  reconstructed later. Some attrition is healthy and
  counting it alongside the rest hides the real figure.

The four cuts take an afternoon with a spreadsheet if the records exist, and the exercise usually resolves a question the leadership team has been discussing in the abstract for months.

The cut people avoid

Attrition by manager is the most informative and the most politically difficult. Businesses cut by team, which is nearly the same thing, and then discuss the team's workload rather than the person running it.

Two cautions, because the finding is easy to misread. A manager who has inherited a struggling team, or who is the first to performance-manage people after years of nobody doing it, will show high attrition for good reasons. And small teams produce unstable percentages — two leavers out of six is not a trend.

But where one manager's team consistently loses people that other teams retain, over more than one year, it is the finding. It will not be fixed by a benefit, a salary review or an engagement initiative, and continuing to treat it as a general retention problem is expensive. What it usually needs is capability support rather than removal — most managers producing this pattern have never been trained, which is the population leadership and people management development exists for.

First-year leavers are a different problem entirely

People who leave within twelve months are commonly counted in the retention figure and discussed as a retention issue. They are not one. By the time someone resigns at month seven, the decision was usually made much earlier, and the causes sit upstream.

• The role was described inaccurately. The most common cause and the most avoidable. What the candidate was told in the interview and what the job turned out to be were different in a way that mattered.
• The onboarding was an administrative process. Documents collected, systems provisioned, and no structured sense of what good looks like in the first ninety days. People who feel unproductive at month three start looking at month four.
• The manager was unavailable. New joiners need disproportionate manager time and frequently get less than established staff, because established staff have louder problems.
• The hiring bar moved under pressure. Roles filled in a rush, against a vague brief, produce a predictable share of mismatches that surface within the year.

Each of those is fixed in recruitment and onboarding, not in retention. A business with high first-year attrition that responds with retention bonuses is paying to keep people who should not have been hired into that role, which is the most expensive available option. The productive route is back through the brief, the shortlist and the first ninety days, which is where recruitment process consulting work usually begins.

What exit interviews are and are not good for

Exit interviews are widely collected and rarely useful, for a structural reason: the person answering has nothing to gain from candour and a reference to protect. The most common stated reason for leaving is a better opportunity, which is true and uninformative.

They become useful in aggregate rather than individually, and only if the questions are specific. "When did you first start thinking about leaving, and what was happening then?" produces better information than asking why someone is leaving, because it locates the moment rather than the justification.

Better still, ask people who stayed. A short conversation with your strong performers about what would make them consider leaving costs nothing and is answered honestly, because the person still has an interest in the answer being acted on. Businesses that do this consistently find out about problems roughly a year earlier than those relying on exits.

The pocket nobody counts: people who have already left internally

There is a category that never appears in the attrition figure and often matters more than the people who resigned: those who have decided to leave and have not gone yet, and those who have stopped trying and are staying anyway.

The second group is the expensive one. A person who has given up on progression, is no longer volunteering for anything, and will do exactly what is asked costs a full salary and delivers a fraction of what they used to. They do not show up in any retention metric, because retention counts bodies. Their managers usually know exactly who they are and rarely raise it, because there is no obvious forum for the observation and no process it feeds into.

Ask managers a different question at the quarterly review: not who might leave, but who has stopped putting themselves forward and when did that start. The answers are consistent, immediately actionable, and typically identify the same underlying causes as the resignations — six to twelve months earlier, while something can still be done. Treating that as part of retention and attrition management rather than as a performance issue is usually the more accurate framing, because the cause is generally structural rather than individual.

Turning it into a decision

1. 1

   Rebuild the number as a table, not a figure

   Rows for teams, columns for tenure bands, with regretted exits marked. One page. The pattern is usually visible without analysis.

2. 2

   Name the two worst pockets

   Not the whole business. Two. Every business has them and generalised retention programmes dilute effort away from where it would work.

3. 3

   Establish the cause before choosing the fix

   Manager, workload, progression, pay, or hiring. These have different remedies and the wrong one wastes a year. An afternoon of conversations with people still in those teams settles it faster than any analysis.

4. 4

   Set a measure you can check in two quarters

   Regretted attrition in those two teams specifically. Company-wide attrition moves too slowly and is too noisy to tell you whether anything worked.

None of this requires sophisticated analytics. It requires the leaving date, the team, the tenure, the last rating and one judgement about whether the exit was regretted — five fields, which most businesses already hold somewhere. Where they are spread across a spreadsheet, a payroll system and someone's memory, that is worth fixing first, because HR reporting and analytics is impossible without it.

The wider point is that attrition is an outcome, not a problem. It is the visible result of decisions about hiring, management capability, workload and progression made months earlier, which is why it responds so poorly to direct intervention and why most retention and attrition management work ends up somewhere other than where it started.