Big data may be the big buzzword for workforce management, but to really understand workforce trends over time, companies need to take a long data view, said Brian Kelly, head of Mercer’s workforce analytics and planning function in Philadelphia.
Long data, a term coined by mathematician and network scientist Samuel Arbesman, takes analytics beyond a “snapshot in time” and gives organizations the historical context necessary to put workforce data into perspective, said Kelly, who joined Mercer in 2011 to help expand the global consulting firm’s workforce analytics capability.
He argues that it is only through this long-term perspective that organizations can gain insights into workforce trends and optimize their human capital investments.
Why is big data analytics important to the talent management practice?
In a knowledge economy, the most important asset a company can deploy is its human capital. Many CEOs and board members today see talent as the differentiating factor in their ability to succeed.
Workforce analytics can help organizations make evidence-based decisions on how to deploy that human capital, and to get a handle on how human capital risk affects their ability to execute business strategy.
What are the biggest challenges talent managers face in taking advantage of workforce analytics?
With big data, you are capturing information from a variety of transactions using advances in new technology. The challenge now is how to make sense out of it.
We use a model, developed by IBM, to break big data into the four V’s: volume, velocity, variety and veracity. HR doesn’t have as much volume or velocity as other business lines. They are not doing millions of transactions on a daily basis.
Instead, what happens in the workforce are trends over time, which is why variety and veracity resonate with HR leaders.
The variety comes from all the systems they use. Whether it’s SAP, Oracle, Taleo, SuccessFactors or some other tool, all of these systems have to talk to each other. Then there is the veracity, which is a huge issue for HR because people don’t trust the numbers.
So when you think about big data for HR, it is all about how you link all of the systems together — variety — and how you can standardize definitions to make the data more reliable — veracity. When you achieve those two things, you can unlock the power of workforce analytics.
Mercer has been talking a lot about long data recently. Can you explain the difference between long data and big data?
We think of long data as having a massive historical sweep, rather than looking at a single piece of data at a point in time. So for example, with big data the head of HR may look at head count at the end of the month, or performance once a year. These are single data points that have no perspective. With long data, you look at all of the employees over their entire life cycle at the organization and see what trends emerge.
How does that work?
Imagine a Facebook time line for every employee that lays out their entire experience at the organization — when they are brought in, every performance review, every compensation adjustment or movement internally in the organization.
This allows you to do a couple of things. Instead of looking at a snapshot in time, long data tells the story of how the workforce is changing. It helps the organization make better decisions around merit adjustments and awards, and it allows true pay for performance because you can understand how that individual is performing.
The level of analysis you are empowered to conduct is also far greater because when you look at workforce data over longer than a 24-month period, trends emerge that you won’t see over shorter transactional reporting.
How can you look at long data for more than one employee at a time?
You take the data for every employee and aggregate it up, then you can begin to query it. For example, you might take all the people who have had the best performance ratings over three years and look at their compensation adjustments, managers they’ve had or their original recruiting source in order to identify trends.
By structuring data in a way that is foundational and aggregating it up, you get a more holistic view. That allows the analysis to happen more rapidly without having to go back and reconstruct the data each time you want to run a new piece of analysis.
How do organizations get started with long data analyses? Is there a specific technology or platform to support it?
Before organizations deploy any technology, they need to figure out what they want to measure, how to measure it and how to link it to organizational strategy. These are foundational aspects of any data initiative.
Then many firms, including Mercer, have developed technology platforms that allow organizations to structure their data in a long data format and they can help you analyze it.
What are the biggest mistakes talent managers make when implementing long data and workforce analytics?
The largest gap we see organizations struggling with is that they fail to educate the end consumers of information on what they are looking at, how to conduct the analysis and how to make decisions using that data.
It’s not enough to put a great tool out there; training also has to occur for folks to start making evidence-based decisions. Too often organizations miss that.
How can the talent management team make the business case for long data?
The arguments that get the attention of the C-suite and the board are all around risk and cost. Executives have certain business strategies moving forward, and the risk to executing these plans have a human capital aspect to them. If you can use analytics to quantify those risks, it will resonate with them.
They also are focused on managing the cost of the workforce and getting maximum productivity. When you can use analytics to optimize deployment of resources, and demonstrate how using long data to make evidence-based decisions around talent can make the business more efficient, they will pay attention.
Sarah Fister Gale is a freelance writer based in the Chicago area. She can be reached at email@example.com.