Some comments on the recent paper from IES on HR Analytics:
I read this new paper which explores some of the subtleties in HR Analytics in a helpful and pragmatic way. I particularly liked the list of example questions that HR Analytics can help with which shows the breadth of opportunity available as well as the challenge for HR people to step into this space.
I did though want to comment on the early part of the paper which seeks to define what “HR Analytics” is. The paper seeks to differentiates it from other uses of HR and other data by saying that to qualify as HR Analytics it must be predictive analytics. I think this is too constraining as it excludes both some useful, advanced analytics and also what many analysts actually spend most of their time doing. This is not to say that prediction is not a good destination to aim for but that analytics does not have to be predictive to count as HR Analytics.
So what is lost if we stick to using the predictive analytics distinction? Exploratory analytics to increase understanding of a situation would be excluded with this apporach; for example, network analysis to understand who communicates with who is advanced analytics but is not predictive. Hypothesis testing to compare two different approaches doesn’t use predictive techniques in the usual sense of the phrase but is a valid aspect of HR Analytics. Prediction may be the end point (in many but not all situations) but using advanced analysis and visualisation techniques to understand a problem area can add a lot of value particularly when used collaboratively with business users. Often prediction may be the long term goal in a problem domain but the current data can only add some insight rather than being predictive; indeed the initial analysis may lead to the need for further data collection or to testing new initiatives before prediction becomes possible. All of these are valid parts of the HR Analytics portfolio. If these advanced but non-predictive approaches are excluded from HR Analytics (as is suggested) but are beyond reporting, we then need to come with a new term to cover these types of analysis! The LV presentation at the IES event last year has an Advanced Analytics stage before Predictive which I think is reasonable and it should be included in the HR Analytics field.
The paper also hints that data manipulation and management is not part of HR Analytics but whilst (predictive) analysis is the interesting bit for the business, the reality is that analysts will spend a lot of time (60-80% in some estimates) cleaning and massaging the data first (particularly when combining sources) and so this inevitably features in many discussions of what HR Analytics is. And it is actually very important since models are more often made better by intelligent and creative use of data (through cleaning and combining data for example) than using clever(er) analytical techniques.
The most important factor in defining HR Analytics is really that it is using and analysing data to solve business problems rather than the technical approaches involved. And since this is driven by the business issues, what is actually done and how it is done will vary from one organisation to another. We should start with the business problem first and decide what the best approach to tackle this is rather than deciding that we must use predictive analytics and applying this to all problems.