Using HR analytics to understand employee 'risk of flight' - learnings from Merck KGaA

Janine Milne Profile picture for user jmilne April 15, 2020
Understanding why and when employees plan to quit can lower costs and boost HR productivity. A use case from Merck KGaA makes the point.

(via Pixabay)

Estimates of how much it costs to replace an employee when they leave an organization can range anywhere from 15% to 40% (or more) of the leaver’s yearly salary, depending on their seniority. Whatever the true percentage, one thing is certain - it’s expensive. So identifying those people who are a potential flight risk, and just as importantly, finding a way to keep them or prevent others leaving, can save organizations big bucks.

It’s a question global science and technology company Merck KGaA is keen to explore. Using  people analytics and planning tech from Visier to reconcile and aggregate HR data about its 57,000 staff across the 66 countries it works in, the company built a model to look at people who leave the company in the first 12 months of tenure. Adam McKinnon, advanced people analytics practitioner at Merck KGaA, explains:

When we started to quantify what that was worth to the company, we felt this was a real opportunity.

The team scanned the entire global organization to find where this trend was more pronounced and decided first to focus on Asia, where between 15% to 20% of new joiners left within the first year.

What they found was that people in junior roles were more likely to leave, which is perhaps no great surprise. What was more surprising was that the time of year people joined seemed to have an effect on their flight risk, notes McKinnon:

If they joined in the latter half of the year, people were more likely to leave within a year. I can’t put my hand on my heart and say this is a definitive answer, but the speculation is that that’s a busy time of the year, we’re wrapping up sales windows and there’s a mad dash to the finish line and it maybe the case that the onboarding process is not as rich or as smooth and not in keeping with what was sold to candidates on the way in.


An HR business partner may have as support 200 people - too many to know them all intimately - Visier data can help them focus on the handful who may be a flight risk, says McKinnon:

We’re really trying to use sophisticated models to synthesize the scale of Merck, so business partners can work effectively because they can’t be everywhere at once.

This kind of project would have been incredibly hard and time consuming to do before the investment in Visier in 2016. Marcel Krauskopf, MerckKGaA’s People Analytics Technology Management Expert, explains:

Back then we didn’t have something like this in place. We had recording systems which were mainly accessible to HR people, but there wasn’t really a system that consolidated the global HR data.

There was the goal to provide something not only to HR but especially to the business that is highly standardized and that also covers security aspects of holding employee data. Of course, HR plays a crucial role when it comes to people management, but the decisions in the end are being made in the business.

It was also a clear way for HR to demonstrate culture change or “data democratization” as Krauskopf calls it, so that about 3,000 line managers in our group can use the data rather than it being the exclusive preserve of top tier 100 executives.

With data from one end to the other of the employee lifecycle, the user cases are unlimited, whether Merck wants to focus on regional or global people issues, believes Krauskopf"

We do have the opportunity to scope into country-specific issues, but what we have always said is we want to stick to globally consistent data. We don’t have data in there which is only for a country. We have the benefit that our HR processes and source systems have been unified within the last 10 years pretty heavily, so we’re not asking 20 different core systems to integrate data.

We have consistent definitions, so when someone in China is looking at attrition, it’s the exact same definition as someone in the US. It’s good you can slice-and-dice it as you want, but it’s also important that not everyone is making their own definition of attrition.


Alongside data integrity, making sure they stayed within the legal framework for personal information in each country was paramount. The other key challenge was to get people to use it. As a science and technology company, Merck’s workforce is already very data-driven which made it a lot easier to encourage managers to adopt the tool, even so, Krauskopf points out:

It took quite a lot of effort to bring it out to market and to make sure that we drove it out consistently across the organization. We will never be up to 100% usage, but we needed to enable key opinion leaders in the organization that could cascade the culture down.

This approach appears to be working well, as they have found that employees were more likely to be a user if their manager was a Visier user. This data-driven HR message has been drilled into the culture and the mantra, says McKinnon, is that there’s no business plan without a people plan.

Becoming more data-driven has also changed the relationship of HR and the business, McKinnon believes.

We have a lot of technically driven people and if I have a meeting with people and I specifically don’t tell them I’m from HR, they just assume I’m from an R&D organization. They are a bit surprised that after we’ve been talking about machine learning and data and combining people data with sales data that when I tell them I’m from HR. I do think there’s a reinvention of what HR means. We are very fortunate to work in an environment where people want data.

Data-led decision making can only get better as more data is fed into Visier, says McKinnon:

We get a lot of data from Visier, it’s clean, it’s reconciled and we can extract that content and run it through sophisticated models and then ideally feed the content back into Visier.

Having the right data at the right time is an invaluable asset, turning a job that could take two weeks down to just a few clicks, but the true value is beyond time saved as Krauskopf explains:

When someone would ask me the value monetarily that this tool provides I couldn't make an estimation of how much effort or time people are reducing, but the real value we are driving here is the culture change.

A grey colored placeholder image