INNOVATE

Unilever Is Ditching Résumés in Favor of Algorithm-Based Sorting

Artificial intelligence and algorithms soon could take control of much of the initial hiring process.

Share on
BY Wanda Thibodeaux - 28 Jun 2017

PHOTO CREDIT: Getty Images

Traditionally, the hiring process has meant plenty of face-to-face interaction with candidates, with managers using interviews to narrow their candidate pools. But technology might give that method of operation the axe, as Kelsey Gee of The Wall Street Journal reports. Dutch-British Unilever, one of the largest consumer goods companies in the world with over 400 brands, is experimenting with a process that relies on artificial intelligence and an algorithm to recruit and sort job applicants.

 

How the process works

 

Mike Clementi, human resource executive with Unilever, explains to Gee that, given how young workers have a strong online presence and use smartphones on a regular basis, it no longer makes sense for the company to continue to recruit only at their handful of go-to college campuses. Unilever's new way of recruiting and sorting applicants works like this:

 

  • Unilever puts targeted ads on career advice sites and Facebook.
  • Individuals click the targeted ads.
  • Unilever redirects those who click on the ads to sites where they could apply for entry-level positions. The company pulls the individuals' LinkedIn profile data to fill out much of the application.
  • Unilever's algorithm scans all the applications that people submit, getting rid of more than half of the candidates.
  • The candidates who make it past the algorithm play a set of 12 short online games, which test factors like short-term memory.
  • Unilever invites the top third (or less) of the candidates who complete the games to submit a video interview on HireVue. The questions focus on how the candidates would respond to business challenges.
  • Final candidates are invited for an in-person interview with Unilever human resources executives and managers.

 

What the AI and algorithms are looking for

 

Unilever has programmed their technology to measure specific elements, such as vocabulary, facial expressions and question response speed. These factors theoretically can reveal clues about the applicants education, intelligence, personality and emotional stability. Unilever thus still gets a concept of whether the applicant would fit in well at the company despite the fact humans don't get involved in the hiring process until the last step.

 

But does it get results?

 

Unilever cautions that more review is needed to figure out if the method actually correlates to stronger employees. But some employees already feel that the caliber of applicants is just as good, if not better, than candidates that managers hand selected. The company also asserts that the overall hiring process is much faster. This has clear benefits for Unilever, but it also can appeal to the applicants, who otherwise might be frustrated at having to complete multiple interviews, and who have been somewhat conditioned by technology to appreciate speed. 4 out of 5 applicants who make it to the final round get a job offer. Unilever can get those applicants from a wider initial pool, as well. Lastly, the technology potentially helps the bottom line twice over--first by saving money in hiring, and second by increasing revenues from establishing a team that is more qualified and likely to work better together.

 

Unilever isn't alone in bringing artificial intelligence and algorithms into the hiring process. Goldman Sachs Group Inc. and Walmart Stores Inc.'s Jet.com both reportedly are using similar options, too, according to Gee. As these companies are able to gather more conclusive data about their results, and as AI and algorithms become even more sophisticated, getting people out of hiring likely will become the new norm for companies who want to stay competitive. The only questions are how companies can smoothly transition into the new processes and when it makes the most sense to do so given their specific circumstances.