By: Greta Roberts, Conference Chair, Predictive Analytics World for Workforce 2016
In anticipation of his upcoming Predictive Analytics World for Workforce conference presentation, Embedding Advanced Analytics into Acquiring, Nurturing and Retaining Talent, we interviewed Vishwa Kolla, AVP, Head of Advanced Analytics at John Hancock Insurance. View the Q-and-A below to see how Vishwa Kolla has incorporated predictive analytics into the workforce of John Hancock Insurance. Also, glimpse what’s in store for the new PAW Workforce conference.
Q: How is a specific line of business / business unit using your predictive decisions? How is your product deployed into operations?
A: My function embeds Advanced Analytics (AA) into every decision making – be it to identify good prospects from the US population, be it to streamline acquisition (i.e. to reduce cycle time from 45 days to 1 day) or to be it to deliver on the promise of customer centricity (i.e., be prescriptive with their needs). Advanced Analytics journey takes about 9 – 12 months to develop, integrate, test, monitor and refine. We follow a 5 step maturity process to embed AA into everything that we do. We started out with one BU (Insurance) and the process transcends BUs and functions.
Q: If HR were 100% ready and the data were available, what would your boldest data science creations do?
A: We will tackle the employee retention problem first. The success of a firm depends heavily on the number, distribution and longevity of star performers. Star performer productivity in a firm increases exponentially with time as they build relationships and learn processes in the firm. After solving the problem of keeping the stars, we will tackle the problem of getting stars in and then follow-up by nurturing them.
Q: When do you think businesses will be ready for "black box" workforce predictive methods, such as Random Forests or Neural Networks?
A: We are ready now. Any PA problem has two dimensions – the inputs (data) and how these inputs are combined (model). Our PA process involves variable selection prior to model build. After this step, we build two classes of models – truly predictive and explanatory. Since the starting variable set is the same, the truly predictive model is used to get most lift and the explanatory models are used to describe their predictive power. We always augment model build with reason codes (for individual scores). These codes are critical to driving model adoption.
Q: Do you have suggestions for data scientists trying to explain the complexity of their work, to those solving workforce challenges?
A: A model is only as useful as its adoption. As data scientists, knowing what the data means, understanding the limitations (data collection, quality, transformation, sufficiency, completeness, modeling, interpretation, action-ability) and finally building visuals to explain these limitations will help with obtaining confidence of Business and Operational users.
Q: What is one specific way in which predictive analytics actively is driving decisions?
A: This aspect is covered in the first question.
Q: How does business culture, including HR, need to evolve to accept the full promise of predictive workforce?
A: It all starts with educating HR and its clients – PA primarily helps with bringing objectivity to decision making. It then starts with being as transparent as possible and giving a look under the hood. It is very important to talk about the limitations and the separation of the signal from the noise. To drive meaningful adoption, it is important for PA function to collaborate and not compete with HR. It is also important to understand that this is a journey and not a one and done deal.
Q: Do you have specific business results you can report?
A: The first step of any effective PA process is to build a business case. This involves a) identifying the business problem (retention say for example), quantifying the cost / benefits (costs of acquisition, productivity losses (direct and indirect), opportunity) and quantifying the impact of intended consequences (how much of this problem can be solved using PA, the costs of false positives and false negatives) and being open to unintended consequences. Depending on the problem, a 15 – 20% ROI should be used as a reasonable threshold when deciding between competing PA projects. The actual dollar impact will be a function of the size of the problem being tackled.
Don't miss Vishwa’s conference presentation, Embedding Advanced Analytics into Acquiring, Nurturing and Retaining Talent, at PAW Workforce, on Monday, April 4, 2016, from 2:40 to 3:25 pm. Click here to register for attendance.