The Three Types of Analytics

Third Entry in the Metrics and Analytics Series

We still have a little bit of work to do before we get into the real nuts and bolts of metrics and analytics.

Last week we discussed the five steps of  analytics and this week we are going to cover the three different types of analytics, Descriptive Analytics, Predictive Analytics, and Causation. Again, I am taking all of this information and summarizing it from the Jac Fitz-Enz book titled The New HR Analytics.

The first type of analytic is the most basic, Descriptive Analytics. This type of analytics takes a look at data and analyzes events that happened in the past to help us understand how to approach the future. With this data, we can look for reasons behind past success and failure. Most management reporting uses this type of data.

As HR professionals, we all know that there are many different groups and subgroups within our workforce. With Descriptive Analytics we can really take a deep dive into our workforce and really analyze each of the different groups. It allows us to discover and describe many types of relationships and differences between the groups and show past and current behavior among the groups. It’s important to remember that this type of analytic is not a prescription or solution, but rather just the exploration of past and current data.

Descriptive analytic models quantify relationships in data into group classifications or characteristic. We can take almost any group characteristic and use it to build descriptive models to understand our workforce. The characteristics can be things such as performance appraisal ratings, skills, education, rank, title, age, etc. Fitz-Enz equates this to what the marketing department does with customer segmentation and calls it workforce segmentation with the purpose of improving the return on investment, or ROI, of our HR services.

The second type is Predictive Analytics. This is where we take data and turn it into valuable and actionable information. It uses data to determine the possible outcome of a future event. It gives us meaning to the current patterns that we see in Descriptive Analytics to identify risks and opportunities in helping us see what might happen in the future.

It will give us a good probability of what might happen in the future. We won’t be able to predict the future with 100% accuracy but we can increase the probability that we will be right by reducing as many variabilities as possible.

A great example in the Fitz-Enz book is building a model for selecting employees to hire, train, and promote for a particular position. The model is based on traits, skills, and experience of your very best employees in that particular position. You will increase the probability of making a successful selection by applying the model to all of the candidates for the job.

The third and final type of analytic is called Causation. This is a combination of Descriptive and Predictive Analytics and is considered the most sophisticated level of human capital analysis.

Causal modeling is used to find the hidden root cause of a problem or to make a business proposition for a human capital project investment. Fitz-Enz cites Dr. Nick Bontis as the most prominent expert of causal modeling and uses an example of the perceived value of a training program by the trainees.

In his model, he wants to determine the trainee’s perceived value of the training program so he adds the perceived value of training by the trainees to their rating of the course materials and delivery.

By surveying the trainees, we can determine the business value of the training program by asking them to rate their perceived value of the training on a scale of 1 to 100, the course materials from 1 to 5, and the delivery from 1 to 5. Throw these into an equation and you get something like this:

Perceived Value Course Materials Delivery Total
.80 X 4 X 5 = 16

And you get a total score of 16 out of a possible maximum of 25. We can now track the training’s effectiveness over time, the value of the training to the organization, etc.

When discussing HR Analytics, we can’t simply rely on the past to predict the future. We have to have knowledge as our base in order to do so. Fitz-Enz puts it this way:

Being able to foretell what is likely to happen with a high degree of probability depends on four things:

  1. Comprehension of past and current events.
  2. Understanding not only trends but also the drivers behind them.
  3. Being able to see patterns of consistency as well as change.
  4. Having tools to describe the probability of something in the future.

HR Analytics takes HR metrics and looks to the future by taking past and current strategic and operational data and adding leading indicators.

We now have a basic understanding of the three types of analytics which will help us understand many of the topics in this series I will cover in future posts and podcasts.

Please note: I reserve the right to delete comments that are offensive or off-topic.

Leave a Reply

Your email address will not be published. Required fields are marked *