The monthly HR report at Capgemini spans about forty pages. Still, HR advisor Angela van den Hoek believed more value could be extracted from it. To demonstrate this, she launched a pilot project analyzing the likelihood of employee turnover.
“Managers sometimes get alarmed by the monthly HR update when there are spikes up or down. But presenting the data differently can already create a calmer picture. For example, by reporting turnover as a twelve-month rolling average, you immediately gain insight into trends. Currently, we only see the pyramid in terms of age, gender, years of service, salary, turnover, and so on. But by linking different data sources, we can uncover relationships we hadn’t seen before,” says Angela van den Hoek.
Angela has worked at Capgemini in Utrecht for thirteen years. In addition to advising managers, she is involved in annual salary rounds and variable compensation. She combined the HR analytics project with a training course from AnalitiQs, which she attended with two colleagues—a people manager/BI consultant and the HR reporting manager. The focus was on turnover among the 2,500 employees in her business unit. “Lately, I’ve been reading more and more about HR Analytics. I’m increasingly convinced that predictions are possible when you have so much data.”
What did the project look like?
What were the objectives of the project?
“I wanted to increase my knowledge of HR Analytics, see what we as an organization could learn from a basic study, and whether we could gain enough support to do more with HR Analytics. To determine the business question, we spoke with several managers. One of them said: in exit interviews, we get an idea of the reasons for leaving; it would be even better if we could predict who will leave. Then we might also prevent unwanted turnover. So we came up with the question: what are the most predictive variables for Capgemini employees leaving? And can we anticipate this with more knowledge?”
How did you combine the HR analytics training and the analysis project?
“The training was divided into four sessions with intervals. Each session began with a theoretical introduction. With this knowledge, we then got to work. The first session’s theme was defining the business question as precisely as possible. We then tested this with colleagues from the business. The second session addressed the question of what data we needed to answer the business question. The third session involved analyzing this data. The fourth and final session was about creating the reports. Irma Doze from AnalitiQs gave us assignments at each training session. She advised us on properly delineating each process step and its execution. My HR Reporting colleague provided the necessary data. Irma then ensured that all the required files were linked to perform the analyses.”
What obstacles did you encounter during the execution?
“One challenge was that different names had been given to categories in files over the years. This made linking the files a very labor-intensive task, especially combining all the monthly files into one file over about three years. Furthermore, data was linked from the personnel administration, a separate database for exit interviews, and the project administration, including absence and leave data. The project administration records which projects an employee has worked on and for what period. A missing element was the project locations, so we had no insight into travel times and distances. Another obstacle was that we couldn’t use the data from the employee satisfaction surveys (MTOs). We had just switched to another research agency, and that transition required all our attention. We definitely want to use the qualitative data from the MTOs in the near future. By collaborating with an external party like AnalitiQs, which also conducts MTOs, we can ensure privacy.”
What results and insights did the project yield?
“We wanted to map out which groups and individual employees are most likely to leave first. We presented this in an overview. The most important variable is the number of years of service. This aligns with what we already knew about reasons for leaving. In that sense, there were no surprising insights. But there were also no insights that contradicted what we already knew. The group with the highest likelihood of leaving has up to six years of service. Then, the likelihood of leaving decreases with the number of years of service. Only in a specific group with longer service does the likelihood increase again. Part of this group goes through a phase where significant changes occur in their private lives, such as starting a family. This can provide reasons for leaving. These results underscore the importance of having good conversations with employees, as this can prevent turnover. We emphasize this in training for people managers. After completing the study, there was enough support to continue with HR Analytics.”
Did you have enough information to fully answer the business question?
“No, to know how to prevent turnover, we need more qualitative information. For example, what makes someone feel connected to the organization and how that plays a role. We hope to gain these insights from the MTO next time.”
Did the project raise new questions?
“A few years ago, we agreed to keep selection assessments for two years. The number of available assessments was therefore limited. Still, examining these assessments already provided an indication of the characteristics that distinguish stayers from leavers. I would like to conduct further research on this once more assessments are available. Additionally, I would like to look more at long-term trends, correlations between different HR topics, and predictive analyses. I will discuss this with my HR director soon. Apart from that, combining the monthly turnover data into a twelve-month rolling average already showed how simple it can sometimes be to make better use of data.”
How do you want to proceed with HR Analytics?
“I would like to combine it with my work as an HR advisor. Being close to the business is essential to properly assess what insights are needed. Colleagues who know the business less well draw different conclusions. For example, a colleague said that turnover is not a problem when we have fewer assignments as an organization. But in a knowledge-intensive organization like Capgemini, it matters a lot who leaves. You want to grow, develop your talented people, and not see them leave. Some people are not easily replaced. Also, to know what data is needed to answer the business question, it is good to be close to the business. As an HR advisor, you have that position.”
Which organizations would you like to take a look at?
“Google. When I see how far they have come and how much they can find out from data, I think: wow! What I like about Google is that they try so many things and constantly adjust their policies.”
Tips for HR professionals and companies starting with HR Analytics
- Involve colleagues from different departments in the organization
- It is important that the research question comes from the business itself
- Ensure recent information and data
- Supplement quantitative data with qualitative data
- Use as much data as possible to avoid missing relevant connections
- Align the final report of the results well with the business question
Pitfalls
- Don’t think that a few days of training will teach you how to tackle HR Analytics
- Knowing what data is available and what you can use requires considerable knowledge
- Without a statistical background, it is difficult to get a grip on HR Analytics
