Recently a disheartening post appeared on LinkedIn in which an HR data scientist was described, who was recruited to produce some great insights, but who was asked instead to generate monthly or weekly reports.
Lyndon Sundmark and Hendrik Feddersen, both long-standing HR professionals and absolutely passionate about HR Analytics, would like to offer their personal recommendations. What follows are some helpful tips written for HR data scientists and their HR counterparts to get an HR Analytics project started.
Helpful tips on how to start an HR Analytics project:
First read widely on the subject of HR measurement and metrics to understand the possible extent of the subject. There are several metrics one can analyse, for example choose one from here or here. It is important to concentrate your attention on a very limited number of metrics. Reading will help you to develop a conceptual framework, which can encourage in developing a starting point of where you want to head. It will help to answer the question of ‘What’ needs to be done.
Consider a framework covering three different categories of metrics:
- HR Activity, i.e. what is going on with the human resources in your organisation and how HR is changing. This is the ‘traditional’ category of HR metrics most often discussed in the literature. And many people give possibly unintentionally the impression that this is all there is. Most organisations start the journey is this category. It’s a good place to start.
- HR Operations: How efficient and effective is HR in providing HR Services? This category is sometimes looked at after the previous category is well in hand.
- HR Methodologies: Those relate to analytics embedded deep in the way HR performs. For example complementing recruitment or job classification with much more quantitative methodologies to assist the decisions made in these areas. Examples of this are using statistical classification methods for job classification or using statistics for predictive validity of tests or data at time of recruitment compared to performance down the road after the hire. This is an important category but is rarely seen in HR, both because of traditional training in HR and often because of the intransigence by the HR community to see the quantitative side of their profession.
Ask yourself the following questions and produce informal prototypes:
1. Why are we doing this? Why begin the journey? Is it the organisation and its executive driving this? Or is this being driven by the HR leaders or by a desire from someone within HR for ‘leading edge’ HR methodologies. Have there been any rudimentary reports done already or are you starting from scratch? Answers to these questions will tell you whether you have external support or you are on your own, whether it will be a formal project or an internal skunkworks project.
2. Do you need ‘permission’ to innovate or improve the way the HR business is done? Or is it expected as part of your role? If it’s expected as part of your role, many improvements start out as ideas and prototypes long before they become formal initiatives.
3. Where is your role ‘vis-á-vis’ HR? Is your role embedded in HR? In almost every organization to have legitimate access to HR data you must be in HR and even then there may be restrictions, i.e. you have no legitimate access to use data contained in the HR information system. Ideally the HR Data Scientists should be embedded in HR to extract HR metrics.
4. Regardless of the category chosen above, take stock of what has already been done with respect to HR metrics, analytics or reporting. This will help to place you in a much better starting point of the HR Analytics effort. Any effort must begin with a starting point and have an ending point. Improving what’s already there may have initially less visibility and risk. (The world existed before HR Analytics came on the scene).
5. At this point you have considered the possible questions of what, why and where. The next major question becomes ‘who’. The answer depends on the ‘what’ especially. What you plan to do determines the skill set needed. HR Analytics initiatives require the integration of three skill sets: HR, Data Warehousing, and Statistical Analysis as per our HR Analytics guru infographics. Do you have all these skills in one person, or are these skills available in HR? If not, co-operation between IT and HR and data scientists will have to occur. Skunkworks projects could have informal collaboration. More formal projects will need more formal teams.
6. Assuming that you do (like most organizations do) start with HR Activity metrics- you should define and document your metrics both narratively and in terms of quantitative formulas. Inevitably the first set of questions asked during presentations is ‘how do you define that’ and ‘what calculation do you use to get that number’. This should be done prior to any technical development of metrics.
7. Another key step before technical development is to ask ‘what questions are likely to be asked on this metric’ and ‘what types of analyses are likely to be needed on this metric’.
Generally two types of needs exist. Either there is a need to understand at any given time what causes a certain metric to be what it is, or there is a need to predict the future, before it occurs, or both. The needs will determine what data is needed, how it needs to be organised, and the resultant analyses needed. You may have to pre-anticipate the questions in the beginning.
8. When you are at the technical development stage, the concentration should be on automating the data extraction, transformation, and load so that the production of the metrics themselves is effortless at the end of each time period, so that the maximum effort can go into the analysis of the metric. This is where your ‘value-add’ is.
If you have the necessary skills and technologies or access to them, it has been our experience that skunkworks projects and/or informal prototypes are a good way to go. They establish a foothold or proof of concept before any formal projects/initiatives takes place. Trying to start a formal initiative and getting approval for it, is a lot like asking permission to invent a car or a smartphone before they were invented. People won’t know enough at the time to give permission.
Thanks for taking the time to read our post. Comments are always welcome.
In our next post we will explain a practical HR Analytics project.
Please follow us on LinkedIn Lyndon Sundmark, Hendrik Feddersen, and/or Twitter to be informed of our next posts and to make us happy.
You are welcome to download our infographics:
HR Analytics Guru
Helpful tips on how to start an HR Analytics project
Keywords: HR, Human Resources, HR Analytics, People Analytics, HR Data Science, Talent Analytics.