The ability to combine business outcome data with HR data when making predictions utilising predictive workforce analytics is one of the most important trends for HR leaders, according to an expert in the field.
HR’s data is largely made up of activity and compliance data, such as when people were hired, their salary, their manager, when they were promoted, their engagement scores and their performance review scores.
However, HR often does not track their actual performance in their role, once they are hired, and this is a very important distinction, said Greta Roberts, CEO of Talent Analytics, Corp.
“Employees are hired to perform a service for the business – sales, call centres, IT, finance and the like,” she said.
“At the end of the day, businesses want to know if employees are performing in their business role. Business performance is tracked in the business.”
To predict actual business impact, predictive analytics models need to include business outcome data and Roberts said this data is easily accessible from the line of business. “Examples include, sales data, call centre data, customer service score data, attrition data, error data, cash drawer errors, and so on,” said Roberts.
“The most effecting predictive workforce analytics include business outcomes.”
Another important trend for HR is in the value of predicting employee flight risk, and potentially intervening to keep a valued employee who might otherwise leave.
“Since the cost of replacing a top employee is staggering this trend makes complete sense,” said Roberts.
“A dilemma with predicting flight risk of current employees is the often awkward conversation a manager has with the employee.”
“Predictive workforce analytics is a lot easier than people imagine”
One alternative is to predict flight risk prior to someone being hired – as a job candidate, Roberts said.
“Imagine two candidates for the same role. They both appear to be great candidates but one has a 27 per cent probability of lasting in their role for one year while the other has a 78 per cent probability of lasting in the role for a year,” she said.
“This flight risk prediction, I’m sure, will add important data to the hiring decision.”
A third key trend for HR is around predicting employee performance for internal and external candidates, according to Roberts, who said finding out an employee is a low performer can be financially devastating.
“In some cases the cost of a bad hire can wipe out the value of three to six top hires,” she said.
“The largest expense tends to be in the on the job training – meaning, the time it takes someone to get up to full productivity which being paid for full productivity.”
Roberts also observed that organisations are finding that data is adding to the talent decision, and not replacing “good old-fashioned expertise” from the employee making the decision.
“This is an important point, and a fear I think, that data and predictions will replace human expertise. This is not the case,” she said.
3 essential metrics
There are a number of essential metrics that should be applied when it comes to predictive workforce analytics, and Roberts said that the most important metric by far is business outcome data.
“The team needs to agree on specifically what they want to predict, and then find the data to support that,” she said.
“Include data that answers the question ‘what performance is happening in the role. This is different than performance review data – we mean actual business performance data.”
Highly targeted data from HR is also essential.
“There is simply no need for HR to have a single unified dataset of all of their data,” she said.
“These kinds of projects are the kind that can make the career of an HR leader”
“Predictive workforce analytics is a lot easier than people imagine. Predictive work does not need not be disruptive; it can use the data you have even if it is in a combination of spreadsheets, multiple systems and the like.”
Roberts also said that predictive analytics “likes a lot of data”, and the most accurate predictions come from being as specific and precise as possible.
“One way to do this is to make predictions in a single role (sales) versus in a single category (level 12 managers),” she said.
“All managers are going to be so different – a sales manager is different from a finance manager. Predictions are possible but they will have a wider error band and will be ‘fuzzier’.”
For greatest accuracy and business ROI, one important metric is to consider finding roles with a large (more than 100) number of people in the same role and creating predictions around flight risk and performance in this role.
“This will increase predictive accuracy and business value significantly,” she said.
Implications of predictive workforce analytics for HR
To win with predictive analytics, Roberts said HR leaders need to focus on solving business problems, not HR problems.
“HR leaders need to locate business units where there are tremendous employee turnover problems or tremendously low performance problems.
“If they are able to solve these problems they will get a quick win, the business will take notice and they will continue to get their analytics projects funded,” said Roberts, who added that these kinds of projects are the kind that can make the career of an HR leader.
“HR leaders don’t need to know how to model, but they must know how to identify the areas in their organisation desperately struggling with turnover or performance challenges.”
Roberts recommended HR leaders follow three steps to get up to speed with predictive analytics:
- Examine where the business has roles which suffer from constant turnover, and the roles in which staff are under-performing and needs more top performers
- Once a pain area in the organisation is identified, then the goal is to find internal or external set of data scientists to help solve the issue
- Once the predictions modelling is complete, the HR leader can make sure that business results are reported to the business leader.
The next issue of Inside HR magazine will feature a special report on predictive workforce analytics for HR leaders. Image source: iStock