A large eCommerce company, delivering a variety of low-unit-value items to the consumer’s doorstep. They have a million customers, and are doubling (both revenue and number of customers) every year. Consequently, they have thousands of employees, all along the Supply Chain (procurement, collection, packing, storing, warehouse management…) and the Delivery Chain (Order Clubbing, Loading, Driving, Delivering at Home…).
Attrition amongst this feet-on-the-street workforce is a major challenge. The higher the attrition, the higher the cost (recruitment, training…) and the lower the efficiency (learning curve and customer satisfaction), employee morale, marketplace reputation…
Senior Management only expects things to get worse as the company grows rapidly. They just can’t figure out WHY employees are quitting.
We carried out secondary research, as well as spoke to experts in HR, with experience in managing large feet-on-the-street workforce. We asked each of them how do we find out WHY employees are quitting. One of them quite surprised us by telling us to stop asking WHY employees are quitting.
Instead, he said we should find out WHO is quitting and WHERE they are going. Once we understand these, the WHY will present itself. By WHO, we were supposed to define the personas or profiles of the employees quitting. Is it a Driver, doing the 2nd shift, in Hyderabad North, employed for less than a year, under Manager XX that is leaving the most frequently? And are they typically going to Bangalore South to work for a Telco managing Retail outlet Sales under Manager YY?
These are the insights that will unlock the answer to WHY. Are people quitting because they don’t like driving in Hyderabad North, or is it because people do not like working with Manager YY? Armed with these details, Recruiters can hire the right people and HR can put in place the right interventions to mitigate the impact of every aspect on the stability of an employee.
The data driven analytics platform caters to these key aspects:
- Collect stated / unstated data points
- Consider external data (macro economic, market research…) to understand ecosystem
- Compute Employee specific metrics to predict probability of churn, degree of influence, etc.
- And provide a one view interface to slice & dice the data
The technical platform leverages different components in Azure – Azure Data Factory for data ingestion, Azure SQL Server and Azure Analysis Services for persisting and processing the data related to reporting and scorecards. Custom machine learning algorithms in Python for churn prediction and forecasting are scored in batch mode and output integrated to Azure SQL server for rendering.
A whole host of immediate (and short term value is being delivered here):
- Created a good insight into the employee pool
- Allows the client to slice and dice the employee pool into different segments, including the segments that has maximum attrition
- This insight goes directly to the HR department, who are working on intervention steps to mitigate attrition
- Given the insight, we should also be able to predict what profile of employee is likely to churn in the next month / quarter / year.
- Provided privacy concerns are addressed, we can also drill down to an individual level
- Similarly, profiles of stable employees can also be created, which will be used by the Recruitment department to source employees
Now the client has greater confidence that they can chase triple-digit growth, without having to worry about attrition!