Co-authored with Peter Douglas.
Want to know what links the top performers in your organisation? They may sell a lot but some of them don’t seem satisfied. Are they paid enough? Are they bored with what they do? If they leave, how easy is it to find a good replacement and at what cost to the business?
Opportunities in HR analytics
A single view of an employee poses a great opportunity to build the kind of analytics capability that can expose subtle trends that help HR to make predictions to inform decision making in recruitment, performance management and retention.
As data science moves into the arena of HR, it’s not just big online businesses like Amazon that can harness the power of their data. Let’s look at some examples of what can be done right now with existing technology to drive down recruitment costs and to help retain and nurture the best business talent.
Advanced and predictive analytics in HR – some examples
There are many use cases relating to the application of advanced analytics to HR, here are just a few examples of what can be done:
Automated segmentation of employees to identify patterns in employment and performance KPIs
Segmentation of employees based on selected KPIs e.g. skills, experience, sales performance, holiday/sick leave, salary/growth, tenure, sentiment, probability of leaving, etc. The identification of cluster groups can provide insight into common factors linking similar employees, for example, see the parallel coordinate chart below. This shows 5 clusters identified in 4000 employee records. Each thin line represents an employee e.g. mid-salary, low tenure employees with mid-range sales have a low sentiment score and high probability of leaving. This alerts the business to a problem area and poses the question as to what can be done to address this.
Resource planning: recruitment and retention
Attrition and recruitment planning based on predicted future resourcing requirements. This could include demand forecasting based on multiple causal influences such as economic factors and industry trends.
New talent screening
The screening of CVs and social media using text mining tools. Scoring and reporting results and ranking of candidates based on these scores. Comparisons between the attributes of prospective candidates and existing employees can be made to determine their ‘fit’ for the job role.
Employee sentiment management
Data mining of emails and survey responses to detect the level of sentiment towards the organisation over time. The detection of sentiment and the management of trends identified can help manage attrition and drive improvements in performance.
Employee risk management
Identification of risks posed to an organisation through the behaviour of an employee. Anomalous behaviour can be identified via outlier detection in large data sets and cluster analysis.
It may seem like the journey to achieving an advanced analytical capability in HR is long and complex, but it needn’t be. Many businesses already have the data in place and are therefore already well on the way to realising this.
The route to achieving an advanced analytics capability in HR is outlined below. It includes a novel architecture playing on the strengths of different technologies.
Technology options and advantages
Technology must be at the very centre of a shift towards the creation of a bleeding-edge analytics solution. Whilst core HR data may exist in one system, the data necessary to do this is often disparate and comes in many different forms. To make analysis of this truly effective it is essential to bring all the data together into a single place. For structured data ‘SAP Data Services’ provides a solution to integrate many disparate sources of information.
Unstructured data may also be valuable when trying to gauge employee sentiment or evaluate the level of risk an employee may pose to an organisation. This data may come in many forms including analysis of Emails and Social networks. The HANA Predictive Analytics Library is available to provide the capability to score this data on entry to the solution, provide feedback for any relevant KPIs and ascertain if any data should be stored, providing truly advanced analytics capability.
As with all analytics projects it is extremely important that a solid foundation is laid ahead of jumping in at the deep end and attempting to solve the more complex issues.
With this in mind the starting point has to be the construction of a single employee view. At this early phase it is essential to consider future use cases for the view and flexibility to be pivotal in its design. It is not necessary at this stage to bring all the differing data sources into the view, these should be prioritised with quick easy wins front loaded against the more challenging sources of data.
Once this baseline has been reached the reporting that HR and other areas of the business require should be visited. Some reporting may already exist however it may be beneficial to move this into the solution and ensure that whatever is produced by various stakeholders is standardised so that the business, in its entirety, is working with a single version of the truth.
The application of more advanced data mining techniques can now be considered in order to start to develop and utilise some of the aforementioned use cases.
In order to start enhancing this solution it may now be advantageous to look beyond the traditional and obvious sources of data. A lot of very powerful information is likely to reside in a relatively unstructured form and this data should be brought into the system to enhance the predictive ability of the current baseline solution.
The application of advanced analytics to HR is becoming ever more essential and could provide a business with a huge return. HR traditionally has lagged behind in this area yet arguably given the expenditure on employee compensation and the impact high performing employees can have, this could be the area which can have a huge impact on overall business performance.
The technology that has been discussed here is relatively complex and provides a cutting edge solution. However this is something that can be worked towards as opposed to achieved in a single implemention. As with any other deployment a detailed roadmap should be created at the start of the project containing a coherent strategy that will ensure excellence in advanced analytics is achieved.