The HR Analytics Management Cycle
I started my research into HR analytics over a year ago. After a careful review of the growing – but limited – existing literature on HR analytics, I bumped into a substantial lack of clarity across HR and management scholarship, and by so-called HR analytics experts in the industry.
There was – and there is to a certain extent – a lack of clarity in relation to subject matter terminology, with a misleading use of AI and data science jargon, whereby terms such as machine learning, algorithms, expert systems, and predictive analytics were used interchangeably.
And more importantly, there was a considerable dose of confusion around the added value that HR analytics can bring to the HR functions and more interestingly, to the business case and strategic role of HR in organizations.
Unpacking the level of confusion, the research objective that seemed obvious was simply to survey the landscape of HR analytics from a multi-stakeholder perspective, including HR professionals and data professionals alike.
But… why is there such a lack of clarity?
The response to that question was pretty straightforward:
Related (free) resource ahead! Continue reading below ↓
51 HR Metrics cheat sheet
Data-driven HR starts by implementing relevant HR metrics. Download the FREE cheat sheet with 51 HR Metrics
HR analytics combines knowledge and expertise from three different disciplines: human resources, data management – also referred to IT systems and infrastructure – and data science and analytics.
Part of this challenge is caused by the fact that human resources management as a discipline has been historically detached from numerically-oriented components and thus, both HR practitioners and academics – with some exceptions – have not developed a quantitative skillset.
This has led both practitioner and academics within this space to ignore or neglect a wider range of knowledge produced in data systems, sciences and analytics, which by the way is key to the HR analytics organizational field.
One of the main contributions in data science is that any analytics cycle should commence with business problem and should end with actionable interventions.
A definition which can be used in our field can read as follows:
“HR Analytics is the process of addressing a strategic HR concern making use of HR data (and business and external data if necessary), encompassing the following components: identification of an HR issue, research design, data management, data analysis, data interpretation and communication, and subsequent action plan and evaluation”.
In this figure we can observe how each of these components is undertaken through a different skillset: HR acumen, HR metrics, Research Methodologies, IT Systems and Data Infrastructure, Data Mining and Interpretation and Convincing Storytelling.
So against this backdrop, I set out to interview nearly 30 participants, mostly Heads of HR Analytics Functions, HR Managers, and HR Analytics professionals. The vast majority of the sample is made up by organizations based in Ireland, some in the UK and some in The Netherlands. Invitations were sent through LinkedIn upon a profile search with the following keywords in their current job title and/or job description ‘HR analytics’, ‘People analytics’, ‘Workforce analytics’ and ‘Talent analytics’. Interviews ranged from 45 min to 1 1/2 h of duration.
A common pattern was clear: a good deal of these organizations have a relatively small HR Analytics teams (3 or 4 team members on average) and were operating across different industries such as IT, aviation, finance and retail.
Most of them multinational corporations, the two most striking features or anomalies if you like were the following:
- the job title of the Heads of HR Analytics was different in each case;
- the professional education and trajectory of these professionals were in some science field (physics, engineering, etc.) or in statistics and mathematics, but rarely in General Business or HR.
When looking into these selected companies in context, we realized that they all shared a common pattern: an HR Analytics Management Cycle. Each stage of the HR Analytics cycle should answer a set of key questions, which will lead the HR Analytics functions towards successful HR analytics projects and solutions (see figure below):
- The identification of an HR concern should be the first port of call, unpacking how much this issue affects the business strategy or any competitive advantage at stake. For this to happen, the HR analytics function along with HR professionals will need to spend time distilling what HR metrics fall under the remit of the specific HR issue they are trying to tackle and how this HR metrics impact on HR strategy or business performance.
- A proper research design should be undertaken. This is something that traces back to traditional research design methodologies. The fact that the organization may use big data and AI-powered analytical tools does not mean that the research design should be highly complicated when building a solid HR model around the HR issue we are solving.
Within the research design, the HR analytics function needs to figure out whether the dependent and independent variables entail HR metrics they have available in-house or whether they will be sourced from one or several databases, or if they need an HR analytics consultant to collect and analyze the data with specific data mining tools such as sentiment analysis or psychometric profiles.
This is also the moment when the team decides what sort of statistical technique should be deployed (e.g. t-tests, lineal regression, logistic regression, structural equation models, social network analysis, etc.).
How future-proof are your HR skills? Take our 10 minute assessment to find out!Start Free Assessment
- Within the data management phase, a new set of skills is needed, namely, the management of computer systems and infrastructure. These skills will facilitate the interrogation of data across databases, management and integration of the data. The engineers will know whether the data can be interrogated and integrated together within the same database to optimize its usability by the HR analysts.A decision would need to be taken as regards using open-source software and infrastructure such as SQL and Python, or whether it’s more effective to deploy integrative platforms such as Alteryx, or simpler analytical and visualization tools such as Tableau or Microsoft Power BI. This is also a good moment to reflect on the possibility of building a long-term data warehouse or ecosystem, particularly if we are handling a strategic HR issue, which will have an on-going effect on organizational performance.
- The data analysis therefore becomes obvious once the data is ready to be analyzed. The data analysis can be carried out in the most suitable statistical package for that case. Statistical packages embrace different levels of complexity, from Excel, to SPSS, RapidMinder, R or Python. Hence, different skills will be deployed depending on the software and on whether the analysis entails any coding.
- Towards the end of the HR analytics cycle, the HR analytics function will combine their statistical skills with the HR acumen amassed by HR professionals over the years. Statistical significance and relative impact should come to the forefront along with a convincing story showing the explanatory power that the results are offering us.This storytelling side of the process is key in order to get the business side on board with the output the HR analytics function has just provided. The last key issue here is to deliver the quantifiable impact of the significant independent variables (e.g. according to our data, if we apply yearly pay increases of 7%, employee engagement goes up by 15%).
- Finally, the HR analytics function along with HR should come up with an actionable plan of interventions in order to finally tackle the HR issue the team has been studying. This will require identifying a number of new recommendations covering potential new rubrics or a set of nudges that can be implemented. This actionable plan should indeed be reviewed to measure the real impact of these interventions and to find out whether they were the most appropriate ones.
The interviews included in this research project uncovered some further interesting discussion points or considerations as regards HR analytics in context:
- Are HR analytics functions reinventing the wheel when it comes to designing HR models and hypotheses development?
There is a tendency to rely on the team’s HR acumen and experience along with practitioner-oriented literature, whereas HR academic scholarship offers a wider range of HR models, which are more robust and offer a more trust-worthy validity. Access to academic scholarship is very limited though and perhaps this is a space where a closer collaboration between industry and academia can provide with some synergies.
- Do we need a more rounded training in HR metrics?
HR metrics are not part of the educational curricula in undergraduate or postgraduate courses in third level institutions in Europe with some few exceptions. HR courses are delivered through one or two-day workshops in universities such as Kings College London, the University of Cambridge, Cornell University and MIT. In this way, online communities and academies such as www.analyticsinhr.com offer specialized, subject matter courses designed along with HR analytics professionals.
- Is it important to take a long-term approach to data infrastructure and ecosystems?
The answer is: it depends. If we have identified HR analytics projects that will enable competitive advantage or will tackle an on-going HR issue, then it is worth dedicating two or three years to build an ecosystem. Even in data-driven organizations that were pioneers in the field of HR analytics have had to spend a couple of years re-tweaking and unifying the code needed for the ecosystem.
Often times, these pieces of code were picked by different developers and in different points in time, so certain standardization was indeed needed, so that the ecosystem’s bare bones could be used and modified by anybody coming after them.
- Do organizations need a large budget to be able to do HR analytics or even to build an ecosystem?
Not necessarily. There is a good deal of software that is rather expensive but there are also open source infrastructure, program languages and statistical software that can be used. Oftentimes, we might need to pay for visualizations tools, or for certain features but the important thing here is that the HR analytics is knowledgeable of the range of options that are available.
- Are visualization tools HR analytics?
They can be depending on the HR issue we are dealing with. Sometimes, we might only need Tableau or Microsoft Power BI but other times we will need to run more sophisticated statistical analysis and these visualization tools oftentimes fall short on that front.
- What is the management challenge in HR analytics?
The management challenge resides on getting professionals from three different disciplines to understand each other’s language. HR professionals, data managers and data scientists speak are in different spaces, and often time there is not a fluid dialogue among them. Heads of HR Analytics need to speak those three languages to a certain extent so that HR analytics projects are carried out in a timely manner.
Two final thoughts based on this research. First, there is still a shortage of HR analytics professionals and capability, particularly in Europe. Second, caution and attention should be given to an ethical use of data: there are elements of HR analytics that can be become problematic as regards: anonymity and privacy, data safety, algorithm transparency, and accountability in the decision-making process.
Are you ready for the future of HR?
Learn modern and relevant HR skills, online