Advanced People Analytics: New Opportunities (I)

People analytics has finally come of age. There is growing interest in applying big data and advanced analytic techniques like ML/AI to people analytics. However, people analytics as a function is still far behind analytics in other functional areas. In this article, I outline a few innovative areas worth exploring and discuss why they would be valuable for the business and the people capital function.

Drive linkages of people metrics and data to customer behavior. These types of analyses can be highly valuable for the following reasons

1.      They provide a very compelling ROI for HR/employee investments. Historically the focus of employee investments has been on employee metrics e.g. attrition or engagement. In my work I have seen that the ROI of employee investments when return is defined as customer behavior can be >10X the ROI of employee investments when return is defined based only on employee metrics

2.      Allows for optimal pairing of employees (especially frontline employees) with customers based on their ability to influence desired outcomes. For example, some employees are better at influencing NPS and others are better at driving cross-sell. In a call center, being able to dynamically pair the right employee with the right customer given their intent can drive better customer experiences

Automate the first-round screening of candidates. This can be done in one of two ways and ideally both are leveraged

1.      Leverage NLP techniques to process resume information. Labeled data on resumes which has the label of whether they were offered a job or not can be leveraged. NLP modeling would involve initially building a language model using techniques like LSTM and then applying the language model to build a classification model

2.      Have candidates take an online questionnaire that assesses various skills. The answers are scored using a pre-built model designed to predict outcomes like getting hired.

Employee level performance index. If we could assign an index to every employee that was a weighted combination of their assessment scores, where the weights were proportional to impact on desired performance, that would be extremely useful in so many ways. First it would allow us to rank employees using a single number as opposed to having to use many different assessments, sometimes in a very subjective fashion. In addition, more importantly these rankings would not be arbitrary but would be reflective of the desired performance in each employee. In addition, we could decompose the index score to determine training and improvement areas. For example, two employees might get the same index score but have completely different drivers of that score which would yield insight into how their training needs to be structured differently

Use non-traditional data to understand drivers of employee behaviors. Non-traditional data would include things like email metadata, calendar data, messenger data, travel behavior, employee intranet post sentiment etc. This can uncover more powerful drivers beyond obvious drivers like incentives and be more predictive. It will help design programs that are more effective in influencing desired employee behavior

In my next post, I will discuss further opportunity areas for advanced analytics along the employee lifecycle