No Average Headcount in Turnover Metric
Period: Here is why!
Hey all,
Two weeks ago, I pissed off a bunch of people with this post where I told LinkedIn that putting average headcount (HC) in the turnover calculation is blasphemy:
There were a few people who agreed, but a lot of people took offence to it and made different types of arguments in favour of the average headcount.
I am stubborn, so I disagreed with them all.
And then, I was like:
What if I am wrong?
In this post,
I review how we got here
My rebuttal of all the arguments
Conclusion
How we got here!
I always calculated turnover as the number of leavers over the HC at the beginning of the period. It comes naturally. It’s the same as customer churn.
Think about it.
Start with some people.
Some will leave (leavers). Some will stay.
Your turnover is the people who left over people you had at the beginning with a conclusion: you have X% of people leave in the period.
Simple. Straightforward. Uncomplicated.
It’s like cookies:
Your rate of eating cookies is 1 per hour
It’s like budget:
Your burn rate is $20 a day
It’s like customer churn:
In a year, you expect to churn 5% of customers
None of these use the ending number because it makes no sense to include the end of the period number in the denominator.
So, why would you do it for employees?
But here are the arguments I have received and my rebuttal:
We use averages to understand deaths, retirements, and promotions.
Just measure them separately.
You don’t need to average start and end.
Wanna know deaths, take the people who died and divide by starting HC.
Retirement? How many people retired in the period over starting HC.
Promotions? People promoted over starting HC.
Note that none of these is really turnover.
We use averages elsewhere and want to be consistent.
You gotta use metrics that make sense.
If we use averages for engagement, it doesn’t mean we should use them across.
Always choose accuracy and precision over consistency.
After all, what if you are consistently doing the wrong thing?!
Averages give us a view into hiring dynamics.
I am not sure how.
A pure turnover is a cleaner metric.
How many people did you start with?
How many have left?
How many do you need to find a person to fill?
These raw metrics give you more info about hiring dynamics!
Use cohort analysis instead of turnover.
There is such a thing as cohort analysis.
But then…
Each cohort has turnover.
They are not mutually exclusive either.
The number can be biased in large organizations going through change.
Yes, rapid hiring, acquisitions, and layoffs happen.
It’s a real thing.
But they are as real as the people who were there in the beginning, and counting how many of these people left.
Versus, how many people who weren’t here left, which the average implies.
If anything, when HC increases at the end, you are biasing your estimates down to be more favourable. But more favourable is not always, more true.
Here is a quick table showing that the raw turnover rate is a more accurate representation of turnover at the company, even in natural growth, promotions, layoffs, and acquisitions.
Conclusion
I still see little argument for including average HC in the denominator.
I think it is a common practice taught in school. But I don’t think it’s right and should be revisited and taught correctly, for that matter (check out my course below).
After all, people analytics is not about applying formulas you learned in school. It is about thinking about what you are trying to understand.
Learn People Analytics in a Practical Way!
Check out my new Practical People Analytics Course that covers the most common questions I get from HR professionals:
What metrics should I use?
How do I measure engagement?
How do I make sure there is no bias in my comp?
What is the best way to measure performance?
How can I use advanced analytics to drive action?
Which means… you will have everything you need to build your data-driven HR function.