Retention Rate

What the retention rate is

The retention rate is the share of the people employed at the start of a period who are still employed at its end, expressed against their number at the start. Its denominator is a group identified person by person on a single date, and its numerator is how many of that same group remain.

Everything turns on that condition: the same group at both ends. The measure does not ask how many people left the organisation. It asks how many remain from a group defined in advance. Someone who joined after the start date is outside the denominator: their staying does not raise the figure and their leaving does not lower it, although they are in the organisation all the same.

One caution belongs at the start. The retention rate and the turnover rate are not two sides of one coin, and neither can be derived from the other by subtracting it from 100. Each has its own denominator and its own numerator, as the calculation below shows.

Retention rate and turnover rate from the same figures

Take an organisation that starts the year with 200 employees. During the year 30 leave and 50 join, so at the end of the year it has 200 minus 30 plus 50, which is 220.

Of the 30 who left, 12 had joined during the same year, so they were never part of the starting group. That leaves 18 leavers from the starting group, and 200 minus 18, or 182, of its members still in post.

• Retention rate: 182 divided by 200, which is 91 percent.
• Average headcount: 200 plus 220, divided by two, which is 210.
• Turnover rate on average headcount: 30 divided by 210, which is 14.3 percent.

Retention and turnover add up to 105.3 percent, not 100. That is not a mistake in the arithmetic. The 91 percent was calculated on 200 and the 14.3 percent on 210, and the complement of retention in this calculation is 9 percent, a third figure equal to neither of them. Anyone who reads 91 percent and then reports turnover as 9 percent has announced a number nobody calculated.

Calculate turnover on the starting headcount alone and it comes out at 30 over 200, or 15 percent. That makes three figures from one set of events, with nothing in the organisation having changed.

The figures here and in the examples that follow are assumed. They are there to show how the denominators relate to one another; they are not a benchmark, and they are not quoted from any source.

What the retention rate and the turnover rate each miss

Now suppose that all 30 leavers came from the starting group and none of the new joiners left. The retention rate falls to 170 over 200, which is 85 percent, while the turnover rate stays at 14.3 percent, because the number of leavers has not changed.

Turnover, in other words, cannot tell an organisation whose established staff are leaving from one that is losing people who joined a few months ago. Retention can. That is why the two figures are read together rather than one at a time: each sees what the other misses. The remaining side of the question, when people usually leave, is measured by average tenure, a third measure and not a version of either of these two.

How the measurement period alone changes the retention rate

A retention rate means nothing without a period attached to it, because retention compounds. Take an organisation whose retention rate is 99 percent in every month of the year, a figure that looks excellent. Its annual retention rate is 0.99 to the power of 12, which is 88.6 percent. That means 11.4 percent of the starting group leaves over the year, and the monthly figure did not say so in any single month.

Most of the mistakes in reporting the retention rate start here: a monthly or quarterly figure appears in an annual report without its period, and is read as annual. The remedy is to write the period into the measure’s title, not into a footnote.

The retention rate of one cohort over several years

The most useful form of the measure follows a single cohort across its years, rather than the whole organisation within one year. Take a cohort of 60 people who joined together:

• After 12 months, 48 of them remain, which is 80 percent.
• After 24 months, 39 remain, which is 65 percent.
• After 36 months, 34 remain, which is 56.7 percent.

The headline reading is that 26 of the 60, a little over two fifths of the cohort, left within three years. The figure a decision rests on is a different one: of those who began each year, how many left during it. In the second year 9 of 48 left, which puts that year’s retention at 81.3 percent. In the third, 5 of 39 left, which puts it at 87.2 percent.

What follows from these figures is that the risk of leaving declines the longer people stay, and that the heaviest part of the loss comes in the first year: 12 of the 26 who left over three years, close to half of them in a third of the period. The first year is where spending belongs, not the later years, in which the cumulative figure looks worse because it is cumulative and not because anything got worse in them.

Voluntary and total retention rate

Go back to the starting group and its 18 leavers, and suppose 11 of them left by their own choice while the employment of the other 7 was ended by a decision of the organisation.

Voluntary retention, which is what remains once the organisation’s own decisions are excluded, is 189 over 200, which is 94.5 percent. Total retention is 91 percent, as calculated above. The gap between the two is 3.5 points, and it measures what the organisation decided for itself, not what happened to it.

Reporting only the total figure makes a deliberate management decision appear in the report as a weakness in the working environment. Reporting only the voluntary figure hides the scale of what the organisation decided. So the two are shown together, with a statement of which one is being compared with the previous period.

The legal rules on how an employment relationship ends are covered in their own guides, such as those on resignation and on the notice period and termination. The definition of the retention rate decides nothing about them.

Measuring the retention rate by segment

A single figure for the whole organisation answers a question on which no decision is taken. The splits that make the figure useful are these:

• Joining cohort. Measuring the retention of the people who joined in a given year across their first years shows the month in which leaving usually happens. This is the use on which decisions about employee onboarding are built.
• Role and unit. A high overall rate alongside a sharp fall in one unit hides the whole problem inside the average.
• Difficulty of replacement. A retention strategy is what sets this split, because the aim is not for everyone to stay but for the people who are hard to replace to stay.

The figure for a cohort 18 months after joining is not set against the figure for a cohort 34 months after joining, because the second has come through its second year and the first has not yet completed it. The sound comparison is made at the same age: each cohort’s retention at month 12, then at month 24, even if that means waiting for the newest cohort to get there.

What counts as a high retention rate varies with the occupation and with the state of the market for it. In the sources we reviewed we found no general figure for a sound retention rate.

What the retention rate gets mistaken for

• The complement of the turnover rate. Retention and turnover are calculated on different denominators, as shown above, so subtracting one from 100 is an operation with no meaning behind it.
• A measure of satisfaction. People may stay because they are satisfied or because they have no alternative, and the figure is the same in both cases. The retention rate says how many stayed; it does not say why those who left did so. For that reason it is read alongside what the exit interview records about reasons for leaving, and alongside the state of the market for those occupations.
• A measure of performance. A high rate in a unit does not show that its people do the work well, and a low rate does not show the opposite. High retention at one particular grade can signal a blockage there rather than stability.

What distorts the retention rate

• Moving the start date. The starting group is fixed on a single date. Whoever changes that date between one period and the next ends up comparing two figures that do not correspond.
• Counting people who were not in the group. This is the most common error in practice, because most reports are built from today’s list of employees rather than the list on the start date.
• Mixing the reasons an employment relationship ended. An employee whose employment the organisation ended is not in the same position as one who left by choice, and putting both in one numerator makes a management decision show up in the figure as weak retention. The effect grows in a period with a hiring freeze or a reduction in headcount, when the figure moves for a reason unrelated to the working environment.
• Ignoring small numbers. A unit of six people that loses one sees its retention fall to 83.3 percent, a figure that cannot be compared with that of a unit of a hundred.

There is also an error of presentation rather than calculation: giving the figure to two decimal places for a small group. A retention rate written as 83.33 percent for six employees suggests a precision the group cannot support, and one more departure takes it to 66.67 percent. In a case like this, state the number remaining out of the size of the group, such as 5 of 6. That count is a more faithful signal than a percentage that moves almost 17 points on a single event.

Before a retention rate goes into a report

The most useful test before a retention rate is reported is to write three things beside it on the same line: the start date on which the group was fixed, the number of people in the group, and the period over which it was measured. Without those three, the figure cannot be compared with an earlier period, with another unit or with another organisation.

Show next to it the number of joiners and leavers in the same period, because retention alone does not say whether the organisation is growing or shrinking, and the same figure means one thing in an organisation that has doubled in size and another in one whose size has held steady. A figure shown on its own gets read as a verdict on the working environment, when in most cases it is the product of two other decisions: who was hired, and whose employment ended and for what reason.

This is an explanation of the concept, not legal advice.

Related terms

• Exit Interview
• Employee Onboarding
• Realistic Job Preview
• Succession Planning
• Cost per Hire