Predictive Analytics in Human Resources

Predictive Analytics are a rapidly upcoming trend in Human Resources. We already know that human resource management systems are gaining popularity due to the workload, it takes off from the HR Department. Be it the performance appraisals, preparing salary slips, statutory reporting, paying taxes, approvals and what not, HRMS takes care of each and every need of HR and Finance department when it comes to Payroll Processing.

Predictive Analytics is now application of Analytics on information gathered by HRMS to predict future outcomes. It is a technology that learns from historical data to predict the uncertain future.

The early adopters of Predictive Analytics are Google and Wikipedia. Actually, the most important instrument of Google’s People Operations is statistics. The interview is fully automated. The questions are computer generated in order to shortlist best candidates.

Most useful applications of predictive analysis till now is in Customer Relationships, Marketing and Stock Market where it yield good results. Predictive analysis is used to study consumer behavior and align marketing campaigns in order to increase sales.

The flight booking website recommend best hotels for young people based upon historical market data is an example of predictive analysis in travel industry.

How Predictive Analytics can be used for Human Resources

HR possesses a massive amount of people data. By applying predictive analytics to this data, HR can take good decisions for employees. However only 8% of the organizations worldwide has this capability as per Deloitte Global Human Trends (2016).

As per Deloitte Global Human Trends (2017) 71 percent of companies see people analytics as a high priority but the progress is slow. The percentage of companies correlating HR data to business outcomes, performing predictive analytics, and developing scorecards barely changed from last year.

Main areas where analytics can be used:

(1)   Recruitment (Suppose 2 years back one company hire employees from specific education background and business performance improves. This can be a good input to hire best talent which analytics software will recommend while hiring for a specific position)

(2)   Performance Measurement (Analytics can be used in performance measurement by using employee past data and forming various data metrics. This will directly align promotions with the business outcome)

(3)   Workforce Planning (Revenue in numbers decreases when two employees from operations team resigns giving some weightage. Over the period this data can be proved useful to plan future actions)

(4)   Retention

A specific HR policy launched this year can be linked to Employee Happiness Index over a period of time, which in turn decreases employee attrition is a sign that in future employees will be more happy if the policy continues.

The key differentiator here is the ability to compare the data over time, across business units or between key groups of employees to the overall organizational outcomes. It is not the standalone metrics that brings the insight, but the ability to quickly build comparisons, identify trends and find outliers that makes the difference

(5)   Pay for Performance

Compensation is directly linked to performance can be achieved by using predictive analytics. One of the best ways to demonstrate this practice is achieved by using “Performance based compensation differential”. This metric expresses how much more high performers are paid compared to their average performing peers.

 Good Examples of Predictive Analytics

(1)   During the highly selective training, the U.S. Special Forces predict which candidates are most likely to succeed. Two key predictors are ‘grit’ and the ability to do more than 80 pushups. Grit was actually a more accurate predictor of training success than IQ.

(2)   Wikipedia can predict who is going to stop contributing to the database. Based upon this, they can deploy policies to retain the contributors by sending thank you mail to them.

(3)   Cornerstone published a study in 2015 where toxic employees (fraud, drugs or alcohol abuse, and sexual harassment) bring down productivity in the organization by 30% to 40%.

The parameters and actions should be flexible while implementing analytics at organization level. As, for each organization the key performance indicators differs and type of actions also differs. The weightage for each parameter will also differ. So, all these should be flexible and configurable from front end itself.

Legal Aspect of Predictive Analytics that must be considered to avoid lawsuits:

(1)   Analytics can’t take into consideration an individual protected characteristics when making a selection decision. Characteristics includes race, gender, age and other protected characteristics that can result in decisions causing disparate treatment.

(2)   Privacy Concerns: Big data initiatives may include sensitive or protected data. Employers must be vigilant to protect employee privacy and comply with the myriad of international, federal, state and even local laws in this area.

Getting Fooled

As pointed out by Eric Siegel, arguably the godfather of predictive analytics,

With so much data at your disposal, you can easily misinterpret the data and drive wrong results which in turn takes your business down. This spree of data exploration must be tamed with strict quality controls.

Analytics may provide you results which differ with the opinion of experienced in the organization so, you must act with caution before taking any decision.

A case of Randomness

People have tendencies to assume trends and actions where randomness is actually the primary factor driving the trends.