Rise of the Machines: Algorithms Make Better Hiring Decisions

Despite the extensive use of cognitive, behavioral, skills, language, and other types of assessments in the hiring process, many contact center recruiters fall back on what they are comfortable with: “this person sounds good on the phone.” All too often this results in hiring employees whose performance is suboptimal and/or are terminated well before they become useful and productive – both of which are very costly.

Predictive analytics – the practice of using advanced statistical techniques from modeling and machine learning to analyze past events and behavior in an effort to predict the future – is one of the leading disruptive business innovations today. It’s not a particularly new concept in the contact center. After all, the time-honored process of forecasting and scheduling is a form of predictive analytics. More recently however, sophisticated predictive analytics software has been developed to identify which applicants to hire based on the characteristics of successful candidates hired before them.

A recent Bloomberg Business article, citing research conducted by the National Bureau of Economic Research (NEBR) has lent some independent validation regarding the effectiveness of this software, making it easier for talent acquisition professionals to embrace these applications. Using predictive analytics early in the recruitment process to identify performance and tenure potential has significant implications as it relates to the selection of the hourly, repetitive workforce, like that in the contact center.

The NBER survey included over 300,000 hires across 15 companies. It concluded that predictive algorithms reliably identified applicants who were more likely to be retained longer and outperformed their human counterparts in this regard. They significantly raise the recruiting team’s organizational value by facilitating better hiring decisions and, consequently, improving the bottom-line performance of the company.

In the contact center context, predictive analytics encompasses collecting as much pre-hire data about a candidate as possible, comparing that against his or her post-hire tenure and performance, and identifying correlations – positive and negative – between the data sets. Using advanced machine learning techniques, predictive algorithms can identify the characteristics of high performance employees and look for those characteristics in subsequent applicants. The collateral benefit of this approach is that it is self-improving over time – as better performing candidates are hired, the bar is effectively raised for the next generation. The cycle continues with each subsequent class hired.

As the leading provider of predictive talent analytics solutions for customer service roles, HireIQ recently analyzed real-world tenure and performance outcomes from several of its contact center clients. Its analysis mirrors the NBER findings – “green” (high potential) candidates perform better than “yellows” (moderate potential), who perform better than “reds” (low potential). Companies that hired candidates who were predicted to be excellent performers (“greens”) enjoyed a 60% improvement in critical 90-day retention; 56% increase in first call resolution attainment; and a 37% lift in customer satisfaction. Those that hired candidates that were predicted to perform poorly (“reds”) saw an increase in attrition and a reduction in performance amongst these employees. This is a substantial endorsement of the real world benefit brought by using predictive analytics in the hiring process.

Does this mean companies can rely exclusively on these algorithms and completely dispense with the human element? Of course not. But it does suggest that adding a predictive analytics solution to the agent selection toolbox will yield significant results and further elevate the recruiting team’s strategic value in the organization.