# How to Reduce Employee Turnover

> Turnover is usually treated as a retention problem and solved with a retention program. Most of it is decided before the person starts, which is why the programs disappoint.

Author: Kariaa Research
Published: 2026-08-23
Category: Managing
Reading time: 9 min read
Canonical URL: https://www.kariaa.com/reports/how-to-reduce-employee-turnover

---

import { RankedList, CompareColumns, Timeline, Takeaways, FAQ } from "@/lib/report-charts";

An operator with high turnover buys a retention program. It runs for a year. Turnover is
roughly where it started, and the conclusion drawn is that the workforce has changed.

The program was aimed at the wrong population. **In hourly work most separations happen in
the first ninety days, and a large share of those in the first thirty, which means the
majority of turnover is generated before anyone had a chance to retain anybody.**

## Split the number before you try to move it

An annual turnover rate averages a population that is not uniform, and the average hides
the thing you can act on.

Someone who leaves at eighteen months and someone who leaves at eleven days are different
events with different causes and different remedies. Blending them produces a figure that
supports no decision. Seasonal separations blend in too, and a fourth-quarter hire who
leaves in January completed the arrangement rather than turning over.

**Report three numbers instead of one: separations inside thirty days, separations between
thirty and ninety days, and separations after ninety.** Almost every operator who does
this for the first time discovers the problem is concentrated somewhere they were not
spending.

<CompareColumns
  left={{
    title: "Leaves inside 90 days",
    items: [
      { label: "The job differed from the description" },
      { label: "Schedule unworkable in practice" },
      { label: "Nobody owned the first week" },
      { label: "Pay after travel not as discussed" },
      { label: "Fixed by hiring and onboarding" },
    ],
  }}
  right={{
    title: "Leaves after 90 days",
    items: [
      { label: "Pay progression stalled" },
      { label: "Supervisor relationship" },
      { label: "No route to a better role" },
      { label: "Life circumstances" },
      { label: "Fixed by management and pay" },
    ],
  }}
  caption="Two different problems, routinely counted as one and solved with one program"
/>

## The early exits are hiring outcomes

Everything in the left column above traces back to something said, or not said, before the
person started.

The schedule was advertised as flexible and turned out to be unpredictable. The role was
described by its clinical or technical half and the administrative half appeared in week
two. The pay was quoted before unpaid travel between sites. Nobody was expecting them on
day one and there was no obvious person to ask.

**None of these are retention failures. They are description failures, and they are fixed
in the job post and the first week**, which is the argument in [Employee Onboarding
Checklist](/reports/employee-onboarding-checklist).

The uncomfortable implication is that accurate job descriptions reduce applications.
Naming a rotating shift, a real pay range and the full scope of the work will produce
fewer applicants and more of the ones who stay. An operator optimizing for application
volume is optimizing for the number they then have to replace.

## The later exits, in order of what they cost

<RankedList
  items={[
    {
      label: "Schedule predictability",
      note: "Consistently the highest-leverage lever in hourly work and cheaper than a raise. A schedule published two weeks out, a fixed rather than rotating pattern, and overtime rules that are known in advance change who can stay at all, particularly for anyone with childcare or a second job.",
    },
    {
      label: "The direct supervisor",
      note: "The most-cited factor in exit interviews across hourly work generally. Also the slowest to change, which is why it is worth naming rather than pretending a program addresses it.",
    },
    {
      label: "A visible route to a better-paid role",
      note: "Where the next role is invisible or informally assigned, the job reads as terminal. A posted, criteria-based path to lead, operator or trainer is among the cheapest retention mechanisms that exists.",
    },
    {
      label: "Pay progression, not starting pay",
      note: "Starting rate is matched by competitors within weeks. A known increase at a known point is harder to copy and gives a reason to still be there in month seven.",
    },
    {
      label: "Being able to swap a shift",
      note: "A worker who cannot move one shift when life demands it resigns instead. A functioning swap process retains people at close to zero cost.",
    },
    {
      label: "Recognition that costs nothing",
      note: "Real and much smaller than the previous five. Worth doing, not worth being the plan.",
    },
  ]}
  caption="Levers for post-ninety-day retention, by effect per unit of cost"
/>

## Ask people on the way out, and earlier

Exit interviews are run late and answered carefully, because the person is leaving and has
nothing to gain from candor.

They are still worth running, with one adjustment: ask when they decided, not why they are
leaving. **The decision usually predates the resignation by weeks, and the thing that
happened at the moment of decision is the actionable one.** The reason given at exit is
frequently the final trigger rather than the cause.

The better instrument is earlier and cheaper. A check-in inside the first two weeks asking
what has been different from what you expected catches the same information while the
person is still there and while the problem is still small.

<Timeline
  events={[
    {
      year: "Job post",
      label: "Where most turnover is decided",
      note: "The schedule, scope and pay described here are the promises that break in week two. Accurate posts produce fewer applicants and more people who stay.",
      highlight: true,
    },
    {
      year: "Pre-start",
      label: "The unowned interval",
      note: "Days of silence between acceptance and first shift. A named owner and one confirmation message removes most first-day losses.",
    },
    {
      year: "Week 1",
      label: "Expected, taught, checked",
      note: "Somebody waiting by name, one person to ask, and a real piece of work done and looked at.",
      highlight: true,
    },
    {
      year: "Week 2",
      label: "The check-in that pays for itself",
      note: "Ask what has been different from what they expected. Fixable on day ten, unfixable on day forty.",
      highlight: true,
    },
    {
      year: "Month 3 onward",
      label: "Schedule, supervisor, progression",
      note: "The population that is left is comparatively stable, and its reasons for leaving are different from the early group's.",
    },
  ]}
  caption="When turnover is actually decided"
/>

## What to measure

Three numbers, none of which require software.

Separations split into the three tenure bands, reported separately from seasonal endings.
The share of leavers who had a check-in inside two weeks. And, for any role you are
refilling repeatedly, the cost of a day of vacancy, since that is what tells you how much
the churn is really costing and it is almost never calculated. That method is in
[Cost per Hire and Time to Fill Explained](/reports/cost-per-hire-and-time-to-fill).

<Takeaways
  items={[
    "Most hourly turnover happens inside ninety days, so retention programs aimed at long-tenured staff are aimed at the people least likely to leave.",
    "Report separations in three tenure bands and keep seasonal endings separate, or the average hides the part you can act on.",
    "Early exits are description failures, not retention failures. They are fixed in the job post and the first week.",
    "An accurate job post produces fewer applicants and more people who stay. Optimizing for application volume optimizes for replacements.",
    "Schedule predictability is the highest-leverage lever after hiring, and it is cheaper than a raise.",
    "In exit interviews ask when they decided, not why they are leaving. The decision usually predates the resignation by weeks.",
  ]}
/>

<FAQ
  items={[
    {
      question: "How do I reduce employee turnover?",
      answer: "Start by splitting the number. Most hourly turnover occurs within ninety days and is caused by the job differing from its description, an unworkable schedule, or a first week nobody owned, all of which are fixed in the job post and onboarding rather than by a retention program. For the population past ninety days, schedule predictability, the direct supervisor, and a visible route to a better-paid role are the levers that matter most.",
    },
    {
      question: "What is a good employee turnover rate?",
      answer: "Comparing against a sector average is less useful than comparing against your own prior period, because site mix and seasonal hiring move the figure enormously. What is worth tracking is the split: separations inside thirty days, between thirty and ninety, and after ninety, with seasonal endings reported separately since a seasonal worker who leaves as planned did not turn over.",
    },
    {
      question: "Does raising pay reduce turnover?",
      answer: "Starting pay is matched by competing employers within weeks, so it buys less than it costs. Pay progression, meaning a known increase at a known point, is harder to copy and gives someone a reason to still be there in month seven. Schedule predictability is generally the higher-leverage lever and is cheaper than either.",
    },
    {
      question: "Why do most new hires leave in the first month?",
      answer: "Because of promises made during hiring that broke during week two: the schedule was described as flexible and proved unpredictable, the role was wider than advertised, the pay did not account for unpaid travel, or nobody was expecting them on day one and there was no obvious person to ask. Each has a cheap fix, and all of them sit in the job post and the first week rather than in a retention program.",
    },
    {
      question: "Are exit interviews worth doing?",
      answer: "Yes, with one adjustment: ask when they decided rather than why they are leaving. The decision usually predates the resignation by several weeks, and whatever happened at that earlier moment is the actionable thing, while the reason given at exit is often just the final trigger. A check-in inside the first two weeks is a better instrument because the person is still there and the problem is still small.",
    },
  ]}
/>

---

*Written by the team at **Kariaa**. Learn more at [www.kariaa.com](https://www.kariaa.com).*