Infinite Possibilities of Analytics in HR

Why is Business Analytics Key to Business Success?
January 25, 2018
Workshop on Sentiment Analytics with R by Ratnakar Pandey - RACE, Bangalore
Workshop on ‘Sentiment Analytics with R’ by Ratnakar Pandey
July 31, 2018

 Foreword

Analytics as a skill set is all pervasive and HR function is not left alone. Starting with creating employee lifecycle data marts to using analytics to drive more insights and performance in each of the HR functions ranging from recruitment to exit, HR professionals are becoming more data savvy. With the objective of reskilling the HR professionals on analytics, REVA Academy of Corporate excellence, an initiative of REVA University, has launched a three months weekend Diploma program in HR Analytics. The program is designed with an experiential pedagogy with practical exposure to case studies, data sets, tools and analytical techniques.

The participants have gone through practical training on Workforce planning, Recruitment analytics, Learning Analytics, Engagement Analytics, Talent Analytics, Job Fitment Predictions, Compensation Analytics, ROI and Metrics in HR interventions and so on. The program culminates in a capstone project where participants identify a business/HR problem and create an analytics solution. The participants chose a range of challenges, attrition/retention prediction, performance drivers, compensation parity, ROI in L&D interventions etc.

This program has helped the participants to see the connections, correlations and even causality between HR metrics and other business measures – all of which can be used to design effective HR strategies. In other words, HR analytics can provide a tangible link between people strategy and the organization’s performance.

We have created this primer to create a knowledge repository of the program showcasing a few live case studies used in the program along with a set of capstone projects implemented by our select participants. There are three parts in this document, one, a glimpse of case studies used in the program, two, a few select Capstone projects deployed by the participants and three, a few endorsements from both the participants and trainers.

We welcome the HR professionals to explore and engage with us in this unique endeavour to learn analytical decision-making skills and enhance your professional skills!

We would be glad to assist you further.

Happy Reading!

Dr. Shinu Abhi,

Director, REVA Academy for Corporate Excellence, REVA University

9972916030 / shinuabhi@reva.edu.in


Part I – Experiential Learning though Real World Case Studies

The three months program in Diploma in HR Analytics uses more than 10 live case studies with datasets to expose the participants with the real world, practical business and HR challenges. Each of these case studies are proprietary prepared by our expert mentors based on their extensive consulting experience in this field. A few are listed here.

Case 1: Operational Workforce Planning and Prediction

Discussion Leader: Sanjay Shelvankar, CEO, ScaleneWorks

Operational Workforce Plan (OWP) facilitates the organization to prepare itself to meet the ever-changing customer needs by increasing the organization’s readiness quotient to service a customer order. This is usually associated with low lead times and optimization of the organization’s cost of carrying ‘Human Talent Inventory’. This proprietary and industry benchmark real-world case study, prepared by ScaleneWorks, helps the HR professionals to ask the right questions to chart an optimal OWP. The process followed are as bellow:

  1. Standardise the current forecasting process
  2. Calculate the current attrition numbers to reach the final gross
  3. Understand the current Workforce Plan to calculate the need to hire from outside vis a vis internal hiring.
  4. Measure the current lead time to reach the optimal sourcing and hiring plan.

Predictive analytics and Forecasting techniques are used to find the optimal plan.


Case II: Renege Prediction and POFU (Post-Offer Follow-Up)

Discussion Leader: Sanjay Shelvankar, CEO, ScaleneWorks

In a typical IT company, 20%-25% of the candidate does not join the companies’ payroll post offer acceptance stage wasting time and money of recruiters and negatively impacting the business and project deliverables. This Ivy IIMB case study prepared in association with ScaleneWorks deals with the following questions:

  1. What are the key drivers that influence the candidate joining/not-joining a company or Line of Business (LOB)?
  2. What rules can be used to predict the acceptance or rejection of an offer?
  3. Devise a predictive algorithm to calculate the probability of acceptance of an offer letter and joining the company after offer acceptance stage.

The analytics solution uses a combination of Logistic Regression, Classification Tree and Neural Networks.


Case III: Job Fitment and Role Prediction

Discussion Leader: Sarita Digumarti, Co-Founder, Jigsaw Academy

Banks typically hire hundreds of graduates every year for the role of probationary officers. Once hired, these POs, go through an intensive training program. Once a participant successfully graduated from the training program, they would then be assigned to a role ranging from client facing, operations, sales, etc. the dilemma of the bank was to assess the most suitable role to a participant which drives maximum performance. This case study is prepared out of a consulting assignment by the Jigsaw Academy team on job fitment and role prediction to maximise the success in that role and in turn to drive a high-performance culture for the bank.

Analytics technique used is Logistic regression along with dash-boarding.


Case IV: Attrition Modelling

Discussion Leader: Manoj Kumar, Head, COE, HR Analytics, HSBC

Attrition prediction is one of the most used cases in HR Analytics. When good performers leave the organization, the business impact negatively in multiple ways, increasing cost and impacting value chain. The case starts with understanding the monetary implications of attrition. The dataset covers employee life cycle data with more than 30 variables. This is a hands-on case which takes a CRISP-DM driven approach to solve a business challenge through data driven analytics. This case study will also bust a set of myths around attrition through scientific hypothesis testing. The predictive model is tested on both train and test datasets.

Hypothesis Testing, Regression models, Decision tree etc are used to build a highly accurate Attrition model.

Other case studies include, sentiment analytics, Compensation Analytics, ROI in Learning, High Performance drivers. We also used a few famous cases, like Project Oxygen, Money Ball etc.


Part II: Capstone Projects by select participants

Project 1: Building a Compensation Parity Model

Together is a financial institution providing a wide range of financial services to the rural poor and low-income households, particularly women. It is registered with the Reserve Bank of India under the NBFC-MFI category.

Current Outreach

Members 1,792,211
Loan Outstanding (Crs) 3,820.63
Branches 418
Employees 5,237

Business Problem

One way to attract and retain top talent is by offering competitive compensation and benefits packages based on the industry benchmarks. As a prominent microfinance company in the country, “Together” has to comply with fixation of salary as per statutory norms while fixing salary structures and at the same time, create competitiveness to attract and retain the best talents. Currently, there was no benchmark study in this direction resulting into perceived unfairness in fixing the salaries resulting in lower morale. This project hence addressed this issue by creating a compensation parity structure by applying data-driven Analytics.

Process

Collate employee compensation and performance data

Develop hypothesis and patterns

Participate in an industry benchmark study

Based on the industry benchmarks, develop a compensation structure

Implement a salary revision with market correction

Build a predictive model for compensation for new employees

Understanding Data

Data collected had, Demographic details, Designations and levels, departments, tenure, joining CTC, increments year on year, current CTC, CTC components, promotions, performance rating etc. Data cleansing was done to remove missing data and outliers. Each role was compared with the market data.

Description of the Data

  • For the purpose of the project, the Operations department with 4875 employees was taken.
  • The company has 6 grades and up to 5 levels in each grade.
  • There are levels with no employees.
  • Current Compensation ranges are not uniform. Min-Max ranges are from 24% to 121% at different levels.
  • Median salaries are above the Market at most of the levels.
Grades No. of Employees Below Min Above Max Within Range
B1 863 856 1 6
B2 3077 999 206 1872
B3 42 0 37 5
B4 0 0 0 0
C1 581 149 116 316
C2 112 5 47 60
C3 50 2 16 32
C4 5 0 3 2
C5 0 0 0 0
D1 78 54 3 21
D2 27 11 3 13
D3 7 2 1 4
D4 14 3 5 6
D5 3 1 0 2
E1 5 0 2 3
E2 5 0 5 0
E3 0 0 0 0

 Data Analysis

At the exploratory data analysis level, min, max and quartile ranges were created for each role. A set of Pivot tables and dashboards with gender wise, role wise, performance wise compensation structure was created to understand the spread. Variance with the market data and the current salary structure were calculated.

Heat map showed significant correlations between gender, tenure, job mapping, type of roles and age with compensation. Two most important drivers were tenure and type of role. A set of hypothesis tests were done to understand the significance of variables in the compensation.

Building a predictive model

Based on the analysis with the current compensation and market benchmark, a median salary was proposed for each role. Further, A one sample t-test showed a significant difference between the actual salary and proposed median. A multiple linear regression models was built with tenure and type mapping to predict the salary with an accuracy level of 89%.

Business Impact

A market correction of overall salary structure based on the newly proposed median would result in better employee morale. The model is deployed on the test data with high accuracy level which can predict any new recruit bringing more transparency and fairness to overall compensation structure.

 Project Leads

  • Marina Alex, Participant, Diploma in HR Analytics, REVA University and DGM-HR, Grameen Koota
  • Geetanjali Bhati, Participant, Diploma in HR Analytics, REVA University and HR, Reporting and MIS, Deutsche Bank

Case II: Does culture impacts performance?

Omega Healthcare[2] is the leading provider of Revenue Cycle Management process and Analytics Solutions in the healthcare industry. Its technology-driven solutions and services, delivered by more than 12,000 qualified employees, help to minimize costs and save time for clients.

Current employee strength

Demographic Mix
Bangalore Employee Strength 4050
Chennai Employee Strength 3010
Trichy Employee Strength 4065
Bhimavaram Employee Strength 283
Hyderabad Employee Strength 15
Manila Employee Strength 570
Cebu Employee Strength 330

Business Problem

Among the many factors that affect an organization’s ability to innovate, compete, and engage employees and customers is corporate culture. Corporate culture is the amalgamation of values, vision, mission, and the day-to-day aspects of communication, interaction, and operational goals that create the organizational atmosphere that pervades the way people work. It’s hard to define and even harder to get right. Since the organization is seeing exponential growth, VP-HR wants to understand how culture and performance of the organization are interlinked. This project hence addressed this problem statement of quantifying culture and applying data-driven Analytics.

Process

  • Define methodology to quantify culture
  • Collect data through employee engagement survey (EES)
  • Consolidating EES survey data and past four quarters performance data
  • Built easy-to-use dashboards and interactive reports that enables anyone to visualize and analyse data with greater speed, efficiency, and understanding
  • Develop hypothesis and patterns to see effect of culture on performance
  • Come up with cultural parameter that is driving performance

Understanding Data

Data collected had, Demographic details, Designations and levels, departments, tenure, survey ratings, cultural index, performance score, team’s attrition percentage etc. Data cleansing was done to remove missing data and outliers.

Data Analysis

  • Hierarchy tree was created to have analysis at each level
  • Simple pivot didn’t work to create appropriate EES and cultural index score
  • Used formulated pivot to arrive at EES and cultural index score
  • Slicers were used to create dashboards with various filters like location, gender, hierarchy level, role, grade etc.
  • Probability of attrition in each team was calculated based on EES survey response
  • RAG (Red, Amber, Green) analysis was conducted to flag areas of improvement in organization
  • Published a dashboard to know the story behind individual team or leader
  • Correlation between EES score, cultural index and Performance

Key findings

  • Teams having low employee engagement score have low cultural index score
  • There is a strong correlation between performance of the team and it’s attrition
  • There is no correlation between cultural index and performance score

Way forward

  • Revisit the methodology to quantify culture
  • Capture behaviours along with qualitative data (Survey)
  • Repeat hypothesis testing

Sample of Dashboard for Team Leads

 Sample of Dashboard for Team Leads

  

 Dashboard for Team Leads

  Project Leads

Gaurav Prakash Baldota, Participant, Diploma in HR Analytics, REVA University and MIS and Technology (HR), Omega HMS

Madhu M., Participant, Diploma in HR Analytics, REVA University and Team Manager – HR MIS, Omega HMS


Case III: Predictive Attrition Model: Using Analytics to predict Employee Attrition

Attrition is a big problem in industries like IT, BPO and KPO etc. 2016 Year end attrition was 21.4% in this sample organization.  Employee attrition is predictable under stable circumstances, wherein a set pattern can be deduced from certain parameters influencing the employee and the organization at all times. However, who is going to leave, when and why, can be answered based on analytical models developed as a result of data analysis.

Business Problem

The main objective of the case study/business is to can help organizations control employee churn through predictive models which can then be used for developing retention strategies.

Process

  • Collected/ collated the active employees and exit employees and their tenure datasets
  • Develop hypothesis and patterns
  • Build a predictive model

Understanding Data

The dataset consisted of 363 observations, 299 Active HC and 64 resigned. The 11 independent variables in the dataset were,

Gender, Marital Status, Age, Education, Tenure in the organization, Monthly Salary, Designation Department, 2015 Performance Rating, Role, Job Grade, Number of companies worked, Job Involvement, Job Satisfaction

Data Analysis

At the exploratory data analysis level, checked for outliers, and using featured selection like a crosstab with chi-square test used to identify coefficients for significance for inclusion or elimination from the model.

Chi-square and Heat map showed significant correlations between all the independent variable and most significant variables were education, tenure in the company, marital status, and department with Attritors vs Non-Attritors. A set of hypothesis tests were done to understand the significance of variables.

 Predictive Attrition Model

  Key findings

Education, Tenure in the company, Marital status, Department were the most significant contribution for attrition.

Building a predictive model

Logistic Regression model included all demographic variables and subsequently eliminated insignificant variables

Based on Decision tree model & Profile Analysis, found Marital status and Tenure were correlated.

  • If the Marital status = Unmarried and Tenure =<4, then 70% will attrite
  • If the Marital status = Unmarried and Tenure is >4 years, then 54.7% don’t attrite
Figure 2: Decision Tree Model
 Decision Tree Model

 Business Impact

  • Since tenure and marital status plays a vital role, scope to review dependent benefits, and long service awards.
  • Scope for Education policy
  • Better rewards and recognition for tenured employees
  • Job Rotation within departments

Project Lead

Theresa Francis, Participant, Diploma in HR Analytics, REVA University and Global HR Lead, GE


Part III: Endorsements

Participants Endorsements

One of the biggest advantage one have in this age of disruption is continuous learning. Three team members and I decided to pursue and understand HR Analytics better and REVA University provided a great environment where we learnt from experts, got to put what we learnt into practice, also to run a live project to understand how we could add value to our businesses. The program REVA has put together is a must have for any HR professional who wants to stay relevant in the coming years. REVA University has also inspired me to continue to learn.

Participants Endorsements - Mr. Winston DeRosario, Head People & Culture, Grant Thornton LLP.

Winston DeRosario, Head People & Culture, Grant Thornton LLP.


Mentors Comments

Mentors Comments - Kiran Bableshwar, Leadership & Capability Development Lead at Accenture Technology, India

“The program has the right blend of theories, experience sharing and tools which make it very immersive and effective. Learners get to discuss real-time challenges and cutting-edge ideas to solve problems. In a nutshell, an excellent initiative from REVA team for HR practitioners to become better at HR analytics!”

Kiran Bableshwar, Leadership & Capability Development Lead – Accenture Technology, India


Mentors Comments - Manoj Kumar, Head, Centre of Excellence, People Analytics, at a Multi-National bank

“Quality of the conversation cannot get better when HR talks the language of correlation, causation, propensity and that too in the context of business outcome – learning quantification, values operationalization, attrition, and the probability of joining, etc.  A few weeks ago, they didn’t know these terminologies and how to apply those in the real-life situation. They challenged themselves. They spent their weekends in learning these techniques. They may not have mastered the each topic, but they have overcome the biggest barrier – “can I do it?”. And they did it. Period.

Congratulation to all the participants of HR Analytics program at RACE, REVA University for facilitating this change.

Manoj Kumar, Head, Centre of Excellence, People Analytics, at a Multi-National bank


Mentors Comments - Sanjay Shelvankar, CEO, ScaleneWorks People Solutions LLP

In my experience very few corporate-centric programs make sense and actually deliver a good ROI on time and money spent. The relevancy stems from the curriculum content and then having facilitators that are expert practitioners themselves, deliver it. REVA-RACE HR Analytics program, I was witness to the immense amount of background work that the institution did through their exhaustive efforts.

I was very happy to meet with the participants from the HR fraternity, who just were so thrilled that they had learnt HR Analytics as a subject and were now able to turn any HR Business problem into a “Data problem” and then find a solution to it. REVA-RACE HR Analytics program I think is a very good case-study of how a corporate-centric upskilling course is conceived and delivered meticulously.  My hearty congratulations and looking forward to see the next batch undergo a much more refined program. Wish you all the very best”

Sanjay Shelvankar

CEO, ScaleneWorks People Solutions LLP


Mentors Comments - Ramesh Soundararajan, Partner, Culstran Consulting LLP

“If you want to look for an immersive, hands on, career altering program on HR Analytics in Bangalore, sign up for the program from REVA University. It covers the basics, has expert practitioners and analytics experts and gives you multiple options to adapt an analytical problem solving approach.”

Ramesh Soundararajan

Partner, Culstran Consulting LLP