Using cognitive automation to solve the age-old problem of hiring the best talent

Posted by Stefan Lint on November 12, 2018.

You cannot follow the news today without reading stories about how robots and machines are taking over the world. One aspect of this revolution is the role of artificial intelligence in deciding who is hired and who is not. Depending on your point of view, this may either strike fear in your heart (or at least creates a level of unease), or may feel like we are finally seeing the promise of AI come true. The reality is more nuanced.

In some ways, nothing has changed. Industrial and organizational psychology has long understood what factors drive job performance and how to assess for these factors, and this has not changed. In that sense some of these recent technologies simply automate what used to be done in pen and paper tests and behavioral interviews. In other ways, these technologies leapfrog what humans can do by processing far more data and finding linkage between data sets.

In the end, this technology holds incredible promise to help organizations better predict who is going to be their next top performer and who might be a dud, no matter how stellar the candidate’s resume and credentials. But in the end, these innovative technology tools are no panacea. In our experience, there are four key factors that any organization needs to get right to be able to implement and use these tools effectively.

Only as good as the data

The outcomes from using these tools are only as good as the data they use. No algorithm can solve for data that is inconsistent, inaccurate, or unavailable. In most large organizations, data about their workforce and its performance exist in distributed systems, making it hard for any tool to connect the dots. Only a few solutions are able to leverage unstructured data.

Performance management data is known in many organizations to be a poor measure of true job performance, especially if performance indicators are poorly defined or understood, and if bell curves and forced raking makes managers consider criteria other than actual performance. Job descriptions, which are another data set that some solutions look at, might not be consistent and not describe the actual job. Many vendors will tout their algorithms but downplay the difficulty organizations might have providing and using the key data ingredient for their algorithm.

Validity of the tool

The US Department of Labor has adopted a set of guidelines1 that describe the criteria that any assessment (tool) needs to meet. These guidelines are particularly concerned with the validity of the tool (i.e., what the tool measures as a predictor of job performance) and the potential risk of adverse impact on any protected groups because of using the assessment. Any assessment should be validated to not have a disproportionate adverse impact and that only job-related criteria are used in assessing and selecting talent.

In selecting any vendor’s tool, it’s important to understand to what degree the tool has been validated for use in job selection, whether the tool has been challenged in court, and, depending of the level of customization, if any additional validation that might be needed to stay out of trouble.

Eek factor

Understanding that some people are not going to be comfortable with these new ways of being hired, some thoughtful consideration should be given to how best to use these tools and how they might affect the experience of the candidates. At the risk of overgeneralizing, it’s likely that Millennials/Gen Y are going to be more comfortable with a video interview or playing an assessment game on their phone or computer. Other applicants might feel a strong need to connect to a human first before spending time on an automated assessment.

Organizational buy-in

Lastly, not all managers will trust that the new tool will recommend the best candidates. Many managers still rely on gut feeling and “I-will-know-it-when-I-see-it” as their modus operandi for selecting new employees. To foster buy-in, it’s important that managers know how these assessments work, how they select or recommend, and on what basis. The dirty little secret of hiring is that unstructured interviews, which are widely used by these same managers, are a poor way to predict future performance, and bias and the halo effect are well-known issues in selection. Ultimately, a well-defined assessment strategy that uses several tools, including structured interviews that assess for job-relevant factors, is the most effective way to predict that your next hire will be a star performer.

Getting it right

Be clear on which problem you are trying to solve

The good news is that there are many vendors out there that can provide a real return on investment, help recruiters spend their time with the most qualified applicants, and help organizations make better hiring decisions. The challenging news is that there are many vendors, and having a clear use case is critical in helping you select the right vendor. Once you have a clear use case, make sure that the data set the tool will need to access exists and is accessible before making your decision.

Confirm that the tool is based on proven science and is validated

Most vendors we meet with understand the science behind employment selection and are allocating resources to ensure (1) their algorithms consider things that are known to predict job performance and (2) are not creating adverse impact in the selection process.

Educate and create buy-in from recruiters and managers

Not everyone will be an immediate convert, and educating both recruiters and hiring managers is critical in reaching the high level of adoption necessary to increase the return on investment for these tools.

Consider the candidate experience

Finally, ensure that how these tools are used in the process enhances the experience for the candidate and aligns with the employer brand promise. Given today’s labor market and the ease of taking to social media to comment on experiences with your organization, you cannot risk applicants dropping out of the process because of a bad user experience or a misaligned expectation of the process.

Innovative technology holds great promise for almost all aspects of organizational performance, and is already delivering in many areas, including hiring. Selecting the right tool(s) and implementing them correctly, alongside the skills human assessors and interviewers bring to the table, can help organizations make the job of hiring easier and faster with better results.

Stefan LintStefan Lint, is a senior manager in Deloitte Consulting LLP’s Human Capital and Workforce Transformation practice, helping organizations solve critical talent acquisition challenges.

1 Uniform Guidelines on Employee Selection Procedures (1978).

One thought on “Using cognitive automation to solve the age-old problem of hiring the best talent

Leave a Reply