16 Descriptive People Analytics
In descriptive analytics, historical data is examined using summarizations, patterns, trends, and relationships to determine what previously occurred. It uses methods like data aggregation, basic statistical analysis, and data visualization to answer the question: “What happened?”
As a People Analytics example, let’s imagine that a company wishes to know how much employee attrition occurred this year. All organizations gather and retain data that can be analyzed to answer this question such as employee headcounts, job details, hire dates, and termination dates, they may also ask exiting employees to select reasons for leaving the company. This data can be used to calculate an annual attrition rate which is a simple type of descriptive analysis that would answer the question. [If you need a quick understanding of the various terms and approaches surrounding employee attrition, here’s a short video where I talk about it: Attrition, Turnover & Retention: What’s the difference? Plus How to Calculate Attrition.]
The analysis could stop with the calculation of an attrition rate and the question will have been answered. But, as we discussed in step 2, successful People Analytics professionals are curious and creative so they usually want to know more. In addition to the rate, descriptive analytics allows you to describe even more of “what happened?” For example, this year’s attrition rate can be compared to prior years to identify changes or trends over time. Or, rates can be compared across different groupings of the organization to identify areas with higher or lower attrition. The length between hire dates and termination dates can be used to describe the typical length of employee tenure. Employee attrition can be categorized by leaving reasons to identify potential areas for improving retention. And, these are just some of many possible examples of how a curious people analytics professional might use descriptive analytics to gain insights about what is happening regarding attrition in the organization.
Descriptive Analytics Skills
To understand what happened using descriptive analytics, focus on building the following skills:
Start with understanding key statistical concepts like measures of central tendency (mean, median, mode), frequencies, percentiles, measures of dispersion (minimum, maximum, variance, standard deviation), and correlation. Many online resources offer free courses and practice problems. Seek out how-to articles and tutorials until you can complete each of these statistical techniques easily in any tool(s) you are comfortable using. Note, while you can learn the math behind each of these statistics and calculate them by hand, I definitely do not recommend that! In People Analytics, you will usually be working with digital files so it can be more efficient to use existing functions that calculate these for you in platforms like Microsoft Excel, Tableau, R, Python, or other available data tools and focus your time on learning which statistics to run and how to interpret them – we’ll talk more about tools and technology later in this chapter.
Basic summary statistics provide a wealth of information to help decision-makers understand and answer key questions about the workforce. Consider a manager who needs to hire a new, additional team member for position x but they’re unsure what the salary should be for the position. Calculating the median salary among all current employees in position x gives the manager a reference point for a typical ‘middle’ salary for that position. The minimum and maximum salaries can be used to identify the range of salaries for that position and can highlight any current outliers. Correlation can show if employees with more work experience tend to have higher salaries, helping the manager decide if requiring more experience is justified for this role. And, once the manager has a potential salary they are considering, they could use percentiles to explain how the new salary might compare to the distribution of current employee salaries. If the manager was considering a salary at the 50th percentile, they now know that amount is greater than 50% of the salaries of all current employees, which may be an important consideration before finalizing the job posting. A manager informed with all these descriptive analytics will be better situated to make a well-informed decision in this scenario. This salary example is just one of many ways understanding what happened can provide insight to support decisions.
People Analytics Stories & Advice
When I moved across the country to take a new job in People Analytics, I knew little about my new company other than what I heard in the job interviews or could find online. So the first thing I did, after completing data privacy and access training, was get my hands on some employee data files. I remember being in a meeting my very first week on the job and replying to someone’s comment with, “Yeah, that makes sense given how many new employees were hired back in 2006.” That person just stared at me like I had some magical powers or was a spy. After a long pause all they could say was, “But, you weren’t here in 2006, how do you know that?”
The data told me. Turns out I did have a special power: descriptive analytics. When scanning the frequencies of employees with long tenures by year, I saw a big spike of employees who all had about the same number of years spent at the company. From that, I figured they likely had all been hired at around the same time and had stayed with the company since. The person from that meeting was impressed and went on to be one of my biggest personal supporters and a champion for data-informed talent decisions in their business unit.
I spent much of my first couple weeks in that job doing as many descriptive statistics as possible. The approach helped me learn a lot about my new company and provided great opportunities to start dialogues with colleagues. When I’d find something noticeably different but couldn’t figure out why, I had an excuse to reach out and ask questions like “What’s with all the employee training hours suddenly jumping up in 2010? Was there a new learning initiative or training compliance requirement?” This helped me build my business acumen in new and interesting ways. It also allowed me to start new relationships with colleagues that were based on informed curiosity about their work in other areas like training and development. Many of my best, most collaborative relationships there were founded on these first data-informed dialogues. In addition to finding interesting data tidbits for myself. I inadvertently discovered a way to acknowledge and recognize the hard work of others – I loved being able to show them that their effort was evident and their impact left behind visible signs in the data.
Note: When I discovered those 2006 hires, I started searching the news and asking a couple of people if there had been an acquisition or other big event in 2006 that might have explained the number. Doing these things was an example of using my data consumer skills to build my business acumen skills so I could be more informed about the analytics of the organization.
Employee attrition is the departure of employees from the organization for any reason (voluntary or involuntary).
Attrition rate is the percentage of employees that leave the organization during a given time period.
Using math to summarize and analyze data.
Finding the "typical" value in a dataset. (Mean = average, median = middle, mode = most frequent)
How many times each value appears. (Like counting how many people gave each response on a survey.)
A percentile is defined as the percentage of values found under the specific values. Think of it as dividing data into 100 equal parts and saying where a value falls (Like saying an employee is in the top 10% of earners.)
How spread out the data is. (Minimum = lowest value, Maximum = highest value, Variance = average squared distance from the mean, standard deviation = square root of variance)
Seeing if two things change together. (Correlation doesn't mean one causes the other!)
Data points that seem way off from the rest.