Talent Matters: Leveraging Machine Learning for Better Recruiting
Analytics offers talent-acquisition leaders and recruiters a better way to find candidates who are a good fit.
By Jean Martin and Aman Alexander
By now, you've probably noticed the very specific and relevant ads popping up on your Facebook newsfeed ever since you "liked" all those fan pages, pictures and status updates. Or you may have noticed that after watching that exciting action-thriller on Netflix, your movie recommendations now contain the best of Schwarzenegger and Stallone. In both cases, these changes are not coming from a clever marketer or analyst, but rather a type of artificial intelligence called machine learning. Machine learning models learn from real-world data (like your movie choices) and use those observations to make sophisticated predictive models.
In the corporate world, machine learning is being used in ever-expanding ways. For example, cybersecurity teams use it to anticipate potential hacking or malware attacks, and investment professionals use it to predict the movement of stock prices. That said, one area where machine learning has yet to reach its full potential is in recruiting, though our analysis shows that the benefits can be substantial.
Not only can machine learning accelerate recruiting processes by automating the process of evaluating candidates, advanced analytics can aid recruiters in identifying suitable candidates they may have otherwise missed. As the competition for quality people heats up, it remains increasingly hard to find and attract good candidates. This is where machine learning can provide recruiters with valuable competitive advantage.
Introducing Machine Learning to the Recruiting Process
According to CEB (now Gartner), 65 percent of applicants do not meet the basic requirements for the position to which they apply. Once in seat, one in five hires is deemed regrettable by hiring managers. Often, this happens because hiring managers lacked the evidence-based tools needed to make data-driven, informed hiring decisions.
Historically, human experience and gut feel were the primary factors hiring and recruiting managers used to estimate a candidates potential for success at an organization. However, what gets you hired is not always what makes you good at your job. As hiring managers and recruiters started to realize this, the use of high-quality, predictive psychometric assessments increased, and so did the quality of hiring decisions. Progressive companies are now taking the next step forward by layering in machine learning to help find the best talent.
One application is the creation of algorithmic assessments for candidate application/resume data. Machine learning is used to analyze millions of resume attributes -- work patterns, prior employers, educational levels, key tasks and responsibilities -- to identify which features and phrases occur disproportionately in historically successful employees at a given company. Of course, "successful" is in the eyes of the beholder and can be defined however the employer chooses. For some it might be employees who stay six months or more or employees who achieve high performance or customer satisfaction scores.
In addition to creating deep data-driven insights, this type of algorithmic assessment takes a mere fraction of a second. Since zero additional effort is being asked of the recruiter or candidate, algorithmic assessments can shorten the time-to-hire by narrowing the pool of applicants instantaneously and allowing recruiters to focus on more value-added tasks.
The insights from this type of approach can be surprising. For example, a recent algorithmic model created for a call center representative role revealed that previous call center experience was actually predictive of poor performance (perhaps due to misalignment of the previous employers training regimens). This runs contrary to common wisdom, and its unlikely that a hiring manager would have expected this without hard supporting data. By precisely linking this employment pattern to success at the company, and determining the exact statistical relationship, this machine learning process essentially super-charges the recruiter's ability to identify these critical patterns and improve their hiring choices.
Machine Learning Minimizes Biases
Critics of machine learning often argue that these approaches simply perpetuate the biases of the human behavior they are modeled upon. The truth is that the cause of bias in algorithmic assessments is not inherent in the use of algorithms, but rather reflect flawed methodologies during algorithm creation.
Organizations will generate the most valuable and objective results if they consider a few critical questions when training the algorithms:
1. With what data am I focusing the algorithm?
2. How can I ensure this model will be representative of my employee/applicant base
3. How can I mitigate the risk of bias or adverse impact on any particular group?
Focusing algorithms on post-hire outcomes, such as career longevity or the likelihood of promotion (as opposed to simply which applicants get hired), removes the potential for mimicking human bias in hiring processes. Using a large historical data set to against which to apply the algorithm ensures the traits of a small, unrepresentative sample are not generalized too broadly. Finally, validating the demographic and gender distribution of algorithmic scores from historical applicant populations can further protect against any potential bias.
When done right, algorithmic assessment can, in fact, reduce bias in hiring processes, at the same time that it improves the efficiency and quality of hiring decisions within an organization. All of this is good news for job candidates, as increased use of objective algorithmic assessment models will open up job opportunities for qualified candidates across the economy who might otherwise have been missed. The results speak for themselves: When applied properly, machine learning can surface many more needles in the ever-growing but ever-competitive hiring haystack.
Jean Martin is the talent solutions architect at CEB. Aman Alexander is product management director at CEB. CEB was recently acquired by research firm Gartner.