In our last post, we covered 4 reasons why candidate screening is not a human task. This post builds on those ideas.
As part of the candidate screening process, some organizations task their recruiters with manual tagging of the resumes they review. The idea here is to increase the usability of talent databases by generating metadata manually. Recruiters will manually tag resumes with skills, capabilities, industries, roles, and any other attribute they value. In theory, this would generate a database with enough depth to become navigable and thusly incredibly valuable to the organization.
Unfortunately, in practice, manual tagging just doesn’t work. Here’s why:
Recruiters currently spend 7.4 seconds reviewing each resume. If tasks like tagging are added to the equation, recruiters now need to do more than just skim. They need to read, interpret, and assign tags. This drastic workflow change will take more time and the only way to scale is to increase the number of people doing this manual review work.
Behavior change is hard. If recruiters have been doing something a certain way for years and a new process is introduced, it’s likely going to be an uphill battle. When some talent records inevitably fall through the cracks, the practice becomes far less valuable because some talent records remain 'dark data' that can't be surfaced easily.
As with any repetitious human task, accuracy will be an issue. People have different interpretations of words and ideas which will indefinitely lead to inconsistent metadata. Biases are a tool that can help humans get the job done quicker, but accuracy and consistency across a group of reviewers will suffer greatly.
A human-driven process also limits the capability to systematically trace the justification for certain decisions. The human decision-making process is complicated and often riddled with subconscious biases that simply cannot be explained. This prohibits the ability to systematically improve the metadata tagging.
These issues all stem from the fact that natural language data is both prevalent in HR and incredibly hard to deal with at scale. Talent acquisition is centered around resumes and greater strategic workforce planning initiatives often involve some form of unstructured language such as employee reviews, descriptions of project work, or written communication.
Whether hiring today or anticipating future organizational needs, it’s imperative to have a deep understanding of your talent. Mission-critical knowledge and insight is hidden in the natural language documents that companies have been storing for decades. The best way to achieve a deeper understanding of the workforce is to unlock the value of this data. With a powerful combination of technologies, this is now possible.
Learn how AI for talent classification can impact your organization in our ebook: Talent Classification Guide for the AI Era of HR.