Enterprises can tap the mountains of data they capture to foster new ways of decision making in hiring
If you are looking to embark on the human resources (HR) analytics journey, look no further than the good old recruitment process.
Once the process starts, recruiters use the recruitment brief and their judgement to shortlist candidates for interviews. Firms typically have huge amounts of historical data on employees who have performed well and those who have not. This clustering of employees can serve as a benchmark in selecting new ones. Top employees outperform average employees by up to eight times, so historical data has great potential.
The interview process has various stages and there are dropouts in every stage. The questions to be asked are: what stage sees the most dropouts and why? Is it due to of low compensation figures, or is the process is taking too long? If low compensation is the reason, what is the delta that will lead to better salary elasticity?
Among those that are hired many drop out in the first six months. The questions here are: what made them quit, and how can the firm identify the kind who are likely to leave soon?
These questions indicate how data-led approaches can help transform the recruitment process. There are opportunities that lead either to process improvements—the domain of efficiency—or to better HR decisions—the domain of effectiveness. An analysis by McKinsey and Co. found recruitment efficiency can be improved up to 80%.
Enterprises can tap the mountains of data they capture to foster new ways of decision making. For instance, the increasing availability of employee data in core IT applications is stimulating a fact-based approach to improve HR decisions.
While the nature of applications of data varies according to enterprises, HR data is mainly being used to describe the state of HR metrics through simple statistical analysis; to construct statistical and machine learning models that lead to predictions and extract drivers that explain the phenomenon being modelled; and to deploy strategies to gauge the efficacy of the extant HR practice at enterprises.
Drivers of HR analytics
Before getting into the nuts and bolts of HR analytics solutions, we should perhaps acknowledge that data and human behaviour could work at cross purposes—just like oil and water. Besides, motivations that drive behaviour are intangible—they can’t be easily quantified.
That said, one can get surprisingly better insights should one have access to longitudinal data. Having longer, time-series data, as it turns out, allows us to have a better stab at the problem. Since the data generation process continues as long as an employee remains in active engagement, access to data per se is not an issue. Moreover, statistical models are adept at identifying changes in employee behaviour over time.
While there are certain HR problems that currently do not have a satisfactory solution within the existing management practices, HR departments can address many of the problems with data-led solutions.
Profiling and segmentation
Segmentation can be performed to isolate groups of employees with similar metrics. Identifying cohorts of ‘similar’ employees reflects common performance drivers, characteristics and priorities. Each of these clusters can be explored with heat-maps and business metrics to derive the underlying narrative of the corresponding segments. The lessons derived from segmentation can be applied to effectively hire new employees and answer questions as to how long is an employee, on average, likely to stay.
Performance targets, especially for large enterprises and publicly-held companies, are so tightly intertwined with the broader stakeholder expectations that any misalignment or fall-back in performance can have considerable material implication. Prediction of the performance of front-line sales people is necessary for forecasting business throughput. Managers can use predictions in creative ways. For example, if the gap between performance and targets is expected to be large, other instruments such as advertisement and promotion or sales incentives can be introduced.
Employee churn is a fairly costly proposition for employers. Losing well-performing employees, acclimatised to the cultural moorings of the enterprise, is a concern. Also, higher employee churn invariably leads to escalation in costs in the form of new hiring, raising compensation levels, increased transactions and loss in productivity.
Predictive models can assist not only in identifying those who are likely to quit but also when they might. Employees who score high on the performance index and high on churn probabilities are the obvious candidates for attrition intervention programmes. With statistical techniques such as ‘Design of Experiments’, the intervention programmes can be optimised to deliverlower employee churn.
Given that filling a new position or a replacement invariably takes time, coupledwith the waiting period between the offer acceptance and joining date, a dropout is bound to cost both time and money. This has a cascading impact especially in delivery-oriented or revenue-generating roles, and offer rejections can at times set back business plans. Predictive solutions can identify specific individual-level likelihood of a dropout, thus allowing the HR team to follow up with suitable actions—resulting in superior conversion rates.
As technology and analytics occupy centre stage in business set-ups, firms slow in adopting technology and building analytics capabilities, risk lagging behind in the market place. The best firms see their employees not only as individuals but also as a rich source of collective data that can be used to make better talent decisions.
Source: Live Mint
Date: 23rd January, 2017