Originally served on February 17, 2020 @ 7:54 pm
Last Updated on July 14, 2021 by Pratik
While HR Analytics is not a new concept, and The Brew keeps dabbling on People Strategy Transformations and Optimizations now and then, but what’s new is multitudes of progress in Data Science, Machine Learning, Ai and Deep Learning. We, at The Brew were excited to try these techniques (specifically R – i personally recommend this as a must for anybody who wants data to reveal new perspectives), spreading its awareness to People Leaders by analyzing public domain data set and marrying the inferences / findings with our People Strategic Solutions. (See all of works & applications of HR Analytics here).
Quoting Bersin, who estimates that more than 65 % of Firms are at Managed Level of process maturity, it equally means a lot of data is getting created, good enough to generate basic insights in to Human Capital Model of any Organization.
Presented here is a very primary, simplistic and yet holistic analytics done by The Brew, The purpose is to make HR and Business folks understand the power of this kind of data, along with sharing how The Brew helps Organizations remodel their strategy and operational agility, towards developing a competitive advantage of what can be their most yielding asset / competence – The Human Capital !!
HR Analytics Dataset
This dataset is publicly available, and is nowhere, even remotely connected to any of our clients. Here is the excel version: Dataset Link !!
This read is structured along the following steps:
- General Idea of Working Environment in this Organisation
- General reasons for Attrition
- Why High Performers are leaving ?
- Primary level make shift solutions to fix the leaks in People Model proactively to improve the retention & satisfaction!!
Inferences First: What’s wrong with the People Model – at an Operational and Structural Level
The Brew strongly feels that this kind of Organisation resembles very closely to an early PCMM Level 2, where things tend to be lot more chaotic, reactive and unrequited complexity & stressed efforts to get the job done. At an Operational level, this is characterized with adhoc & inconsistent decision making in Capacity planning, Work Allocation, Performance Ratings, Promotions and Rewards & Recognition.
At a System & Structural Level, The Brew detects that the organization is suffering from the following:
- Lack of Systems, Defined & Standardized Processes and possibly a, total business minded leadership missing the aim to create a resilient sustainable organization.
- The People Value Chain is characterized by 3 things – of what is present, a lot of it is broken or dysfunctional and rest of it is missing!! The disparities in Utilization, positive correlation between Last Evaluation, Average Monthly Hours & Number of Projects indicate the broken unaligned levers of Workforce Optimization, Talent Management & Gratification, whereas low correlation between Promotion & Evaluation denotes a broken Progression, Rewards & Recognition mechanisms. The missing pieces are related to Development, Career planning, Engagement & Enablement, which is definitely hurting the People Model, but ideally should be implemented for fixing the broken segments.
- The Lack of singular enterprise level fabrics of Culture, Competence evident from missing correlation between Experiences in Company to Satisfaction is the bigger piece in the missing jigsaw !!
Areas of Growth: Possible Transformation of the People Model to take the Enterprise to the next level
This should be another blog, to cover it in details, however, in the current scenario, The Brew does advise the People’s Office of this Org to do following in the very short term, not dispensing that is just a make shift solution to keep system moving:
- Fixing the leak of Business Productivity Loss and Risk by giving an understanding of why high performers are leaving.
- Fixing the basic of what’s present of People value chain here, which can improve the longevity, satisfaction & business metrics directly – Capacity Planning & Utilization, Workforce Optimization, Progression & Rewards and Recognition mechanisms.
Getting back to Data & it’s dissection
Comprising of data points from 15,000 employees, the variables covered are Satisfaction level, Latest evaluation, Number of Projects worked on, Average Monthly Hours, Experience in Company (in years), Work accident, Promotion within the past 5 years, Function and Salary. Apart from this, The Brew does not know time frequencies of the data, Geo spreads or any of demographic or socio-economic parameters.
- Number of Projects
- Average Monthly Hours
- Experience in Company
- Work Accident
- Promotion in Last 5 Years
- Last Evaluation
- Satisfaction Level (should have been in Lag Indicators, however due to missing data points on other People Perceptional indexes, this will be taken as a Lead Indicator)
General Idea of Working Environment in this Organisation
The dataset has:
- About 15,000 employee observations and 10 variables.
- The company had a turnover rate of about 24%
- Mean satisfaction of employees is 0.61
Present herewith are different statistical measures of central tendency and variation. For example we can see that our attrition rate is equal to 24%, the satisfaction level is around 62%, performance average is around 71%. We see that on average people work on 3 to 4 projects a year and about 200 hours per months.
Even this simplistic summarization conveys a treasure trove of information for us to understand the People modalities in the given dataset.
Being a People Leader, 4 most important variables here, apart from Attrition are Satisfaction Level, Last Evaluation, Experience in Company, Salary. These variables has the potential to affect the Top Line and Bottom Line directly, while other variables like Number of Projects, Average Monthly Hours, Promotion in Last 5 Years are People Variables influencing the workforce at Individual Levels. The very Critical variable of Work Accident, cannot be quantified further than the times of its Occurrences times, with People’s Office having limited influence to it’s prevention, deterrence or treatment post occurrence. Hence this analysis will limit further deep dives in this direction.
Points of Interests
- Consider the Q1, Degree of Closeness between Mean & Median & Q3 for Satisfaction Level, Last Evaluation, Experience in Company & Salary classifications
- Similarly, extend the same analysis for other variables and one starts to get the operational aspect of People Model at this Organization.
Inferences from this data point
- Performance (Last Evaluation): 25 % of the workforce is operating at around 55 % of Desired Levels (Alarming), Mean and Median are quite close, indicating a spread similar to Standard Normal Curve, which is like, 50 % of the organization straddles around a rating of 0.71 (Concerning) & the Top 25 % Performers are more than 0.87 (Satisfactorily Appeasing), however the disparity between the bucket is a concern.
- Satisfaction: 25 % of workforce is at 44 % (Alarming), Median > Mean, curve seems to be slightly left skewed, indicating around 50 % of workforce being moderately satisfied and 25 % of workforce have approval rating above 0.82. Note that, if NPS is applied here, than only 13 % of Workforce is a Promoter, rest being passive or detractors!!
- Utilization (Average Monthly Hours): 25 % of the workforce is underutilized by 25 % at least. At the same time, 25 % of the workforce is over utilized by atleast 23 % – with remaining 50 % at average 200 Hours !! Aint a healthy picture by any standards, as a good upper cap for average monthly hours would be around 180 hours – but this far exceeds the benchmarks.
- Tenurity (Experience in Company): Looks a pretty young company, with Youngest tenured 25 % of people in organization being 3 year old and Oldest tenured 25 % of People being more than 4 years old. The Mean is greater than Median, indicating the right skewness of the curve, meaning around 50 % of workforce between 3 – 3.5 years.
- Similar readings can be made from Number of Projects & Promotion in Last 5 Years, with the undertones of mismanagement and skewed utilization of resources & a broken Progressions & Rewards and Recognition mechanisms.
Correlation between all the Lead Indicators for Entire Workforce
Few interesting insights from correlation visualization:
- The Positive correlation between Last Evaluation, Number of Projects & Average Monthly Hours. This is a worrying trend, as workforce tends to be rated higher with the number of projects they are in, and equally tend to put in more hours as number of projects increases.
- A positive in this environment is Allocation of projects and Promotion are not dependent on the Experience in organization. This is a great cue for hard selling merit, however lack of correlation here does not mean that the system is designed in such a way, it might equally mean a total ignorance or blinding away from these variables.
- Check out the row of Experience in organization, and there are no clear signals coming out – this is a worry trend as we can figure out why Attrition is happening, but we still wont be able to figure out why Tenured People are sticking to this environment and their motivations!! Effectively, a long term People Strategy is missing and needs to be shaped out for Workforce aided by different levers in People Value chain to give them a strong reason to stick to the organization.
- Check out the Satisfaction row & one finds the no brainer here – Negative correlation between Satisfaction & Attrition, meaning workforce with lower satisfaction levels have higher probability to leave the organization.
Insights from Lead Indicators for entire workforce
Let’s put few parameters on Distribution plots: (Satisfaction – Evaluation – Average Monthly Hours – Experience – Number of Projects)
- Satisfaction – There is a huge spike for employees with low satisfaction and high satisfaction.
- Evaluation – There is a bimodal distribution of employees for low evaluations (less than 0.6) and high evaluations (more than 0.8)
- Average Monthly Hours – There is another bimodal distribution of employees with lower and higher average monthly hours (less than 150 hours & more than 250 hours)
- The evaluation and average monthly hour graphs both share a similar distribution.
- Employees with lower average monthly hours were evaluated less and vice versa, as validated from the correlation matrix, where Last Evaluation, Average Monthly Hours & Number of Projects are positively correlated.
Overview of summary – Averages of variable for Attrition group & Non – Attrition group
Notice that apart from Satisfaction and Salary (not present in above table), there is not much difference in the variables for the people who have stayed and for the ones who are attriting. This is the single most reason why Satisfaction is included as Lead Indicator for deriving analysis here.
Insights from Lead Indicators of Attrited Workforce
Let’s put few parameters on Distribution plots: (Satisfaction – Evaluation – Average Monthly Hours – Experience – Salary)
The U shape formations above are apparent on why people are leaving and why we don’t want to retain everybody. Some people don’t work well as we can see from their evaluation, but clearly there are also many good workers that leave. Equally, check the graphs below – Tenure spreads and Salary
Understanding Salary variable for People Staying v.s. Attrition
- Majority of employees who left either had low or medium salary.
- Barely any employees left with high salary
- Employees with low to medium salaries are at a higher risk of attrition.
Attrition Function wise
- The Sales, Technical, and Support were the top 3 functions to contribute highest to Attrition
- The Management & RnD function’s share was minimum
Attrition & Project Count
- More than half of the employees with 2,6, and 7 projects left the company
- Majority of the employees who did not leave the company had 3,4, and 5 projects. The sweet spot for number of projects allocated seems to be between 3 – 4, as well resonating the undertone of employees not utilizing their potential to the fullest
- All of the employees with 7 projects left the company
- There is an increased risk of Attrition as project count increases
Attrition & Evaluation
- There is a bimodal distribution for people attriting.
- Employees with low performance are a higher attrition risk.
- Employees with high performance are as well, at a high attrition risk
- The sweet spot for employees that stayed is within 0.6-0.8 evaluation
Attrition & Average Monthly Hours
- Another bi-modal distribution.
- Employees who had less hours of work (~150hours or less) are at a higher attrition risk.
- Employees who had too many hours of work (~250 or more) as well are at a higher attrition risk.
- Employees who left generally were underworked or overworked.
Attrition & Satisfaction
- There is a tri-modal distribution for Attriting employees
- Employees who had really low satisfaction levels (0.2 or less) are at a higher attrition risk.
- Similarly, Employees who had low satisfaction levels (0.3~0.5), are at a higher attrition risk.
- And so are Employees who had really high satisfaction levels (0.7 or more), at a higher attrition risk.
Project Count & Monthly Hours
- As project count increased, so did average monthly hours
- Notice the fuzziness about the boxplot graph, where there is difference in Average Monthly Hours between people who left and those who stayed.
- Employees who did not attrite had consistent Average Monthly Hours, despite the increase in projects.
- In contrast, employees who did have a turnover had an increase in Average Monthly Hours with the increase in projects.
- As well, employees who left worked more hours than employees who didn’t, even with the same project count.
Project Count & Evaluation
- There is an increase in evaluation for employees who did more projects within the Attriting group.
- For the Non-Attriting group, employees had a consistent evaluation score despite the increase in project counts.
Satisfaction & Evaluation
- In this visual, 3 distinct clusters for employees who attrited emerge; and this shall form a part of Primary People Management plan recommended earlier.
- Cluster 1 – Top Left (High Performers & Least Satisfied): Satisfaction was below 0.2 and evaluations were greater than 0.75. Which could be a good indication that employees who left the company were good workers but felt horrible at their job. (These as well includes the Overworked Employees)
- Cluster 2 – Centre Focused (Under Performers & Moderately Satisfied): Satisfaction between about 0.35~0.45 and evaluations below ~0.58. These are employees who got low evaluations and in a way are an anomaly, as they left the organization. In a place where they could have managed to stay further, they attrited – may be they were wrongly rated and they found a better offer coming by their side. We need more datasets to probe this further.
- Cluster 3 – Top Right (High Performers & Satisfied Employee): Satisfaction between 0.7~1.0 and evaluations were greater than 0.8. Which could mean that employees in this cluster were “ideal”. They are satisfied with the company and were evaluated highly for their performance.
Why High Performers are leaving the workplace in context?
Post digging the general reasons for attrition earlier, let’s go deeper to understand all the variables that affected High Performers to attrite i.e Bad Attrition. For quantification & classification of Bad Attrition, let’s assume these as base values –
- Evaluation: 0.75 or;
- Experience in Company: 3.5 or;
- Number of Projects: 5, Now, let’s understand how the lead indicators are correlated for this segment:
On an average, valuable employees that attrite are not satisfied, work on many projects, spend many hours in the company each month, aren’t promoted and majority of them are paid low as compared to their peers!! Now, that’s some serious introspection to do for the People’s Office of this organization.
So where does one focus to get better outcomes in the existing situation ??
The Brew recommends following actions, in the order of priority to realize the better of existing human capital:
- Fix the leak of High Performers on High Attrition risk (the correlation matrix above & the workforce which represent those characteristics) on an urgent basis with rationalization of projects from High Performers to Medium Performers, followed helping them with a realistic picture on Promotion and / or Salary enhancements.
- Optimize the workforce with lowest 20 % of Evaluation, that’s not adding any value, adding to non – performance culture and to cost. Improve the basic standards of compliance & org ambiance for the workforce with lowest 20 % Utilization to up the higher realization of work & value, equally setting right tones for culture of accountability & responsibility.
- Rationalize & Standardize the basic fundamentals of People Value Chain with Objective thresholds on Utilization, Accepted Performance Standards, Unbiased Objective Performance Evaluations, Progressions & Rewards and Recognition levers. That should get this organization to be at par with PCMM Level 2 !!