No other metric is more fundamental to Human Resources (HR) than head count. It’s used in so many other calculations that obtaining an accurate count is vitally important for other more interesting analytical metrics. Head count is simple in its idea which is count up everyone that is an employee of the company. Counting is easy right? We can agree on how to count since we were like five years old. Counting is a universal truth. Well, in a word no. What we count and how we count it is very much up for debate, and it can radically change the outcome depending on our assumptions. We are going to explore those assumptions and discuss all of the ins and outs surrounding this simple metric.

The When in Head Count

It is common for someone to ask “What is your head count?” But that’s missing is what point in time do you mean? Normally when asked this question we mean right now, but when you analyzing historical data head count changes over time. What your head count is at the end of the year isn’t the same as at the beginning of the year, or this quarter vs. last quarter. It is very much a point in time metric. Head count makes no sense without time.

Generally people track it month to month. But, it could change week to week or day to day even. Today Tom is an employee, but tomorrow he is not. It changes during that month so when do you mean? Technically you could talk about it as a day to day metric, but seldom is that necessary. But the problem still exists which day do you pick out the month? What does it mean to say December’s head count is 2534?

Typically the last day of the month is used as a proxy for the month’s head count. Sometimes people choose the 1st day or the middle of the month, but the last makes a lot of sense in that the it is who finished the month as an employee. The following equation makes sense using the last day of the month:

[Head Count of prior month] - [terminations of this month] + [New hires of this month] = [Head Count of this month]

The Period

Talking about Head Count month to month is relatively easy in that it is the count the last day of the month. Terminations and new hires can easily be summed up within a single month without needing to adjust anything.

However, what does it mean for head count of longer time periods? For example, what is the quarterly or yearly head count under these situations? It gets a little harder to define those. This is where you can interpret it differently across longer time periods. Generally there are two ways to handle longer periods.

  • End of the period
  • Average Head Count

End of the Period Head Count

Much like we had to decide how to define head count for the month by picking the last day of the month to measure it. For longer periods we can easily pick the last month of the period to be the head count of that period. For example if we are talking about Q1 by month might look like the following:

End of Period Head Count Example

So for Q1 it would be March’s head count of 2710 since March is the last month of Q1. For Q2, June’s count would be used. If we were talking about year period it would be December’s count. The idea behind using end of the period is that we want to represent all of the employees who were still employed at the end of the period.


The benefits of this means that head count is nice round number (this could change depending on other factors), but if we’re counting whole people it will be a nice integer. It also only counts the people who made it to the end of the period and will continue to be employees into the next period. The same mathematical relationship discussed above holds true for EOP Head Count.

[Head Count of Prior Period] - [Terminations of prior period] + [New Hires of prior period] = [Head count of current period]

This is a very nice relationship to have since it makes it easy to interpret the data and check that what you are looking at is correct. In a real system there are a few more metrics added to make it work in all situations, but this is the simplified rule without getting into the weeds.

Trade Offs

The downside with EOP is it ignores what happens between the beginning and end of the period. Anyone who contributed during the period, but did not make it to the end of the period is ignored. This results in it underestimating contribution. If you have high turn over EOP will underestimate this.

Where this matters is when head count is used in other metrics which can really make a difference. For example, Revenue Per Employee can be affected by people who termed within the period because you paid them prior to terminating and up until their termination they must have had a contribution to that revenue. Therefore, using end of period head count in Revenue Per Employee will over estimate the Revenue Per Employee if you have lots of terminations within a period.

Average Head Count

The other algorithm is to combine the counts across all of the months by taking the average across each month within the period. So for Q1 we’d average the head count of Jan, Feb, and Mar. For a visual example of that think of the following:


From a mathematics standpoint it feels better that we are not throwing away data within the period. It is more accurate in our example average head count is 2741.7 vs 2710 for end of period head count. Using an average also captures the people that may have been temporary where using EOP ignores them entirely. So if your workforce has high turnover or uses a lot of temporary workers this will help you understand that relationship, and this might be very important when creating other metrics that depend on head count. For example, Revenue Per Head is just as understandable if you use EOP or Average.

Trade Offs

The obvious trade off is this introduces the problem that we might end up with fractional people. What does it mean that there was 0.7 of a person? It is strange to see the fractional number for head count, but one way to think about this is that 2741 worked the full period and 1 person worked 70% of the period. This is a nice way to explain it so it’s more understandable. However, that is not exactly true since you might have 10 people terminated in a period where their fractional amounts will add up to whole numbers leaving just 1 or a hand full of people that only work 70% of the period.

Because we are not counting individual people anymore there is another big downside. Our nice mathematical relationships between periods does not hold anymore. This makes it hard to verify period to period the correctness of our numbers. Average Head Count does not map back to actual whole people, and that is confusing for people to internalize. It is an approximation of people’s contributions; not an actual count of people.

Full Time and Part Time

Unfortunately head count complexities do not stop with time. Not all employees work the same amount of time every day or week. So should a person working 20 hours a week be counted the same as someone working 40 or 60 hours a week? Generally, a full time employee working 40 hrs / week will be counted as 1. Therefore, a part time employee working 20 hrs / week will be counted as 0.5 since 20 hrs / 40 hrs = 0.5. This is where even using EOP head count could become fractional because of part time workers.

This introduces the concept of a new metric called full time equivalent (i.e. FTE). A value of 1 means they are counted 1 for 1 as you would expect. For a part time person they would be an FTE of 0.5. Head Count is then just adding all of the FTE metrics together.

Normally this metric doesn’t change frequently or respond to the actual hours an employee works. For example, someone who is scheduled to work 40 hours, but works 55 hours one week and 45 the next. This metric works off the scheduled hours for that employee; not the actual hours. In theory it could do that to be more accurate for workers that are paid by the hour, but that requires having the time card data to calculate the actual hours worked.

Why is this so hard?

Head count is often used as a substitute to indicate effort. The more people involved the bigger the effort, and head count is central to comparing populations of differing size. To get two groups on equal footing we divide a performance metric (say revenue) by the people involved so we can compare performance outcomes. This is why it matters so much because these comparisons in analytics happen all the time.

The realization that not all employees count the same amount begins to break away from the simple concept of counting actual heads. We are trying to represent people not as physical people anymore, but their relative contribution to the company or performance. This is where fractional people all begin, but realize that counting in this manner is not counting people, but their contributions. The way in which we think about head count when we shift to contribution changes the meaning of it. It’s not actual people anymore, and counting part time as 0.5 and average head counts changes that meaning.


Head count is one of the most basic metrics for HR analytics. It is ubiquitous showing up in almost every other metric like termination percentage, head count growth, or revenue per employee. Getting an accurate representation comes down to what you are analyzing and how your organization is structured. The mixture of employee types and what metrics you are using can be influenced simply by how you want to count employees.

The difference is counting employees or measuring their contribution is a subtle distinction often confused by practioners. This can lead to under estimating or over estimating depending on the algorithm choosen. Hopefully this has given you some insight into this often overlooked metric and how you can use these algorithms in your company.

Exit mobile version