Turnover: Why DO We Divide by the Average?

Turnover always commands the most attention in HR analytics. Yet, even for daily HR analytics practitioners, there seems to be a bit of mystery surrounding the measure itself….that average in the bottom part. 

The Basic Turnover Rate Formula

In numerous other posts, we talked about the basic method:

\frac{ Total\ Employees\ Leaving\ in\ Period }{\left( \frac{\# \ Employees\ at\ Period\ Beginning\ +\ \# \ Employees\ at\ Period\ End}{2} \right)}&

More compactly, we can just say that we are dividing by the average of the number of employees at the beginning and end of the periods:

\frac{ Total\ Employees\ Leaving\ in\ Period }{Average\ #\ of\ Employees\ at\ the\ Beginning\ and\ End\ of\ Period}&

But WHY the Average?

If you wondered about this, you are not alone. I probably get asked about this more than anything else and I wondered about it too the first time I came across the standard turnover measure.

By averaging the number of employees at the beginning and the end of the period, you get an employee headcount that effectively captures the true workforce size over the period of interest given the natural, constant inflow and outflow of people.

Why This Works

I like to think of it this way. The point of the turnover measure is to know what proportion of your people left in a give time period.

So, ok, first count those who left. Simple enough.

Then you need to divide by the number of people working for you in that period.

But wait! At what point in the period? Who was actually working for us and when?

In particular, what about the new people we hire in the period? Do they count?

If we lost any new hires, we should also include them in the numerator.

But if we include them in the numerator, then we need to include them in the denominator as well.

But should we count them as much as people who were there for the entire period given that they probably joined at some random point in middle of the period?

Similar questions pop up for those who left. Should we count them as “full” employees for that entire period even though most of them probably left at some random point in the middle of the period too?

The answer to all of these questions is to simplify and calculate, on average, how many people were working for us at any point in the period.

To get this average headcount then, we just average the number of people at the start and the end and just use that as the denominator.

Final Thoughts

If you want to get stats oriented, think about it like a stochastic (i.e. random) process in which some people are added (new hires) and other people subtracted (voluntary turnover) randomly throughout the period. Generally then, your best estimate for how many people where working for you at any given time during the period would be the average of the number of people at the start and the end of the period.

Technically I suppose you could calculate how many people where employed for each day individually and then average each of those individual daily measures to come up with a more precise number. But is that really worth all the trouble?

Absolutely not.

This is not chemistry or physics. These little differences don’t have any practical impact on the use of the turnover measure. If we can use analytics to guide is in the right direction, the big stuff will be taken care of.

Think about it. Would you really take different courses of action if most of the people left in the first week of the month rather than the last week?

You care about big numbers and trends, not differences of hundreths of a percent. If you work for a small company, then you probably care less about any given metric and more about the fact the Lisa or Antoine left and now you now have a glaring hole to fill.

I hope this additional explanation answers any lingering questions you have about calculating turnover….but if not, just drop me an email at john@hranalytics101.com

As always, thanks for reading HR Analytics 101 and keep the questions coming.

Contact Us

Yes, I would like to receive newsletters from HR Analytics 101.