As an HR Professional, there are several questions on a day-to-day basis that you may have to respond to.

Most of the questions are with regards to the attrition rate of the organization. While HR is not only concerned with attrition, there are several scenarios, where if HR professionals are able to give an idea about the overall attrition rate, it becomes easier for them as well as for the management team to budget and plan accordingly.

One should understand the fact that old-traditional ways of understanding and calculating attrition rates are long gone and now there is a requirement for more data-backed explanations about attrition.

What are organizations doing wrong?
Most of the organizations are still trapped in the same old attrition formula, which they are modifying as per their convenience with regards to averaging as per quarter or half-yearly basis, to get an attrition percentage.

Some managers can be clever enough to take create an attrition report department-wise, which shows that they have worked on the data and are trying to draw some insights.

Let's understand what kind of insights usually we get from our attrition data;

1. How many people have left our organization?
2. What is the overall attrition percentage?
3. Where is our workforce going?
4. What was our average attrition?
5. What is our quarterly attrition?
6. Why people are leaving?
7. What is average attrition as per department?
8. What are the percentages of reasons for attrition?

etc.

This is essential to know that all the questions that we see above fall in Descriptive Analytics, which means that we are actually only trying to understand what is happening in the organization. 

We have some past data and we are creating Pie Charts and word clouds to depict department-wise attrition and the reason for attrition.

While methods like these help us to understand what has happened and why it might have happened, we are always left blind to things that might happen in the future.

As an HR, we have to move from operation's partner to Strategic Partner, and to accomplish that, we should be able to understand the reasons for attrition and then use that data to predict what will be the attrition in coming months or to be more specific, who is going to leave our organization in next few months.

How to predict if someone is going to leave the organization?
Now, as we have already understood the why of this all, we can move forward with predicting and changing our role to more of a strategic partnership with the management.

But before you actually start predicting, check if you have the resources to do that. Many times, HR professionals have the vision to predict something like that, but they don't have the data or budget to use the tools.

Below are some of the steps that will take you a bit closer to how to go ahead with the prediction of specific attrition;
1. Check if you have the Required Tools to take this project further: As you move towards predicting the attrition of an organization, please check if you have the required historical data.

In several small organizations, HR's don't have access to data with regards to salary data, as it is processed by the Accounts team. They might even not have access to some of the employee benefits data.

This is important to ensure that you have a hold of the data that is required to take this project further.

2. Start Maintaining Data: Data from the start of the employee lifecycle is required for this project. You may require data with regards to the following things to analyze who is going to leave the organization;

• Source of Recruitment
• Psychometric Testing Reports of Employees
• Aptitude Reports
• Education Qualification Details
• Work Experience Details
• Demography of Employees
• Geography of Employees
• Family Background
• Health Details
• Credit Score
• Onboarding Process
• Attendance Report
• Training & Development Reports
• Salary and Incentives Data
• Employee Benefits (Tangible/Intangible) Details
• Performance Management Report
• Official Chat Group Data
• Exit Interviews
• Alumni Data

These are more or less all the details that we can gather to continue further and understand how to predict. If you don't have any of the data, start collecting the same.

The reason for this is that we require as much information about an employee as possible to feed the same in our algorithms, which will help us get an attrition prediction score for each employee.

Data like Psychometric Tests or Aptitude Tests can help you get an insight into the personality traits of employees, which in turn you can use to check if the employee is fit for the organization or they have a personality type, which has been more in people who were leaving us.

3. Understand the Data: It is essential to go through the data before analyzing the same because you need to know what are the various data points and what analytical method you can use to do the predictions.

For example, you can use psychometric tests to compare new joiners with people who are already working in our organization.

This is important because we can not only use one kind of analytical technique on all the data that is collected. Psychometric Test comparisons may use grade systems to compare employees or new candidates, while Chat Data might require sentiment analysis to get insights.

So, it is better to first understand what data sets we have and then go further to analyze.

4. Implement tools required to collect and analyze HR data: There can be several tools that we can implement to collect and analyze HR Data. Several HR Tech companies are providing a one-stop-shop for all kinds of HR Dashboards, Employee Self-Service Portals, and Data Analytics Tools.

Some of the companies that you can consider but are not limited to are;

• Keka
• Darwin Box
• Paysaral
• Salesforce (yes, it can be used too)
• Simply HR
• Orange HRM
• thewiki Network

These organizations provide assistance with the implementation of already built tools or even customized tools to some extent, that can be used for collecting data and then analyzing the same.

5. Have an HR Team: Many organizations when they start have few people working with them, therefore having a small HR Team is usually the best thing that they can do. But as organizations grow, scaling up the HR Team is also important, both in terms of technology as well as people working there.

You have so many tasks to complete. Although you might have automated most of the tasks, you still require a team to look after HR Operations, so that you have more time to analyze the data that you have.

6. Check for Open Source Software: When we talk about analysis, we should understand that most of the analysis can be done using MS Excel and MS Word. But when we are talking about data where we have 100 data sets about 10,000 employees, it becomes impossible to hold the data in excel let alone do the analysis part.

Ensure that you have invested in some tool to handle HR Data or use some open-source tools for the same. Some of the examples of opensource software to analyze HR Data;

1. RStudio
2. Python
3. Power BI (Free One)
4. Tableau
5. Monkey Learn (It's a website for Data Analysis)

I have been teaching some students at an institute that was founded in the early 1970s. HR Students want to gain more practical knowledge about analyzing HR Data rather than learning theoretical concepts.

You should not be afraid to use this software and websites, because most of them are just fearful with the names. You can easily learn them with few weeks of training.

7. Train Your Staff and Yourself: It is important to get trained first and then invest resources in analyzing the data. This is because most of the things in the HR Analytics sphere are still developing and we have to keep up with the changing scenarios.

And you don't want to invest in some software, which employees are not able to use also.

These were some of the points that you must take into consideration while you are moving towards predicting the attrition rate.

Fun Fact
Most people think attrition rate can only be predicted using historical numbers. That's wrong, because numbers will only give you one-sided view of how much attrition will be there, those are not considering some important internal or external factors like policy changes, or the pandemic.

Let's try to predict who is going to leave our organization now,
Now as we have gathered all information and invested in tools and training, we can start with predicting which employees might leave us next, hence predicting the overall attrition rate.

Remember that we have gathered information with regards to several internal and external factors as well. We are going to use the same to predict.

Let's follow basic steps to achieve the prediction analysis;

1. Compare psychometric tests of new candidates and our employees: We can get some idea about what kind of personality traits are with us for a longer period of time.
2. Check the demography of employees: This is again important to know because we will be able to know if we are an equal opportunity employer or we just brag about the same. Are females leaving our organizations sooner? This might be an indication that our new hires, which are females might be the next to leave. There can be several reasons behind that, might be our work culture is not that efficient in cultivating equal opportunities.
3. Check if there is some change in external factors: Several external factors like change in government policies might affect a set of people. For example, when there were other allowances clubbed for calculation of PF, many people left organizations, because they were inefficient in communicating the same to employees.
4. Analyze official chat groups: It is important to keep track of the Sentiment Score of official chat groups. Once you have implemented some new policy, closely analyze the sentiment change and figure out of someone is being more negative towards a particular policy or work environment.
5. Have anonymous feedback analysis: When people know that they will not be found out, they tend to give more honest feedback. We can use that data to analyze if someone is going to leave a particular department.
6. Establish Connection with all the above datasets: It is important to establish a connection with the above data points because you can then only come to a point effectively when you take these all things together.

Try to understand, why the sentiment of the organization is leaning towards a more negative side or why some personality traits are not gelling up.

Are some personality types a mismatch to our work culture?

These kinds of questions will help you understand what is happening and which person is getting the highest score on the prediction scale.

That person will be the one with the highest leaving chances.

How will it benefit us?
Try to image an organization, where you are in India and your people are working from all around the world. 

Think about how can this method help you not only in getting an idea about who is going to leave, but why they are going to leave as well. 

We are not learning this method to figure out which wicket is going to fall. It's unethical to do that, but we want to focus on reasons, why it might happen, and how we can stop that attrition from happening.