Hiring Trend: Machine Learning Reduces Bias, Increases Applicant Pool


(Image: Vicky 81/iStockphoto)

(Image: Vicky 81/iStockphoto)

The Unilever brand is associated with things like soap and deodorant. But technology does play a central role in its business operations, and the UK-based company sought to tap into its power to improve hiring. With the goal of diversifying its pool of entry-level candidates, Unilever experimented with replacing its standard approach to recruiting with algorithms and targeted mobile ads.

As described in a recent Wall Street Journal article, Unilever's recruiting in the past had centered primarily around eight college campuses and the usual resume collection. But the company wanted to try a different approach that would reach more people and filter through applications without overwhelming the human resources in place.

The solution was setting up a hiring process based on algorithms. Based on the assumption that the recent graduates are digital natives, Unilever capitalized on the internet as a both a recruitment and assessment tool.

To capture the attention of potential applicants, Unilever posted ads on Facebook and sites for people seeking guidance on careers. The company also made use of LinkedIn, allowing applicants to draw on their profiles there to quickly fill out applications.

That approach yielded close to 300,000 applicants, which was far too many for the people at Unilever to read through. The company's algorithm cut that number in half, scanning for potential matches for the job.

Next, Unilever put applicants through a dozen online games to test for abilities Unilever wants for the job. Then, the process asked the top 30% of applicants to participate in a video interview where they were asked to answer questions about what they would do in particular situations.

AI was then applied to the video responses, checking for response times, as well as verbal ability, and even the visual cues of the candidates. The video interview further culled the applicant pool, eliminating about 80%.

The remaining applicants went on to meet with an actual person. The process cut out a great deal of human-level screening, increasing efficiency. According to Unilever, The Wall Street Journal reports, according to Unilever, that "Eighty percent of applicants who make it to the final round now get job offers, and a similar number accepts."

The new process has also helped Unilever expand its hiring base substantially. In the past applicants for jobs in the US and Canada hailed from about 850 schools. The new process pulled applicants from 2,600 different schools. They were not, however, all represented in the actual hires, which amounted to about 200 positions for the US and Canada, according to the figures in the Wall Street Journal report.

Andy McAllister, a Unilever director of supply chain, was impressed with the results, saying that the candidates selected by algorithm were "as strong, or stronger" as those "he had hand-selected the prior year," according to the report.

McAllister considers the approach to hold "tremendous" potential for improving recruiting and, possibility, reducing the effect of bias that can influence hiring. The bias can result from the natural inclination for people to identify with particular candidates and favor those candidates. The algorithm eliminates that, though it can't guarantee a bias-free experience.

While Unilever's algorithm-driven approach is innovative, it is not altogether unique in our digital world. The Wall Street Journal reports that other organizations including Goldman Sachs Group and Walmart Stores' Jet.com are using similar digital tools in recruitment.

Businesses are making these moves based on research into the advantages of taking a more data-driven approach to hiring. Plenty has been written about the approach. In October 2016, Harvard Business Review published How to Hire with Algorithms. It referenced a an article from the American Economic Review published in May 2016 called Productivity and Selection of Human Capital with Machine Learning that demonstrated the gains of the machine learning approach to hiring based on studies of hiring police officers and teachers.

Now with Unilever's experience, there is additional data on the efficiencies of hiring for the corporate context. It makes sense from the business perspective to increase the number of applicants and then filter them down to interview only the most qualified.

In the past, businesses often paid external recruiters between 20% and 30% of a yearly salary for just that service. But with the advance of algorithms, that kind of job will remain as relevant as buggy-whip manufacturer in the age of automobiles.

Ariella Brown,

Ariella Brown is a social media consultant, editor, and freelance writer who frequently writes about the application of technology to business. She holds a PhD in English from the City University of New York. Her Twitter handle is @AriellaBrown.

Hiring Trend: Machine Learning Reduces Bias, Increases Applicant Pool

Unilever is using machine learning tools to increase the pool of entry level applicants and then test and evaluates candidates in the pool to cull those who are the best fits. Adding machine learning tools reduces bias, increases diversity, and makes the whole process more efficient, according to Unilever.

Analytics and School Attendance: A Laundry Story

Appliance maker Whirlpool suspected there was a correlation between student attendance and access to clean clothes. Here's the story of how the company placed washers and dryers in schools and tracked the difference it made in student performance.


Re: Subjectivity and Algorithm's Can't Play Nice.....
  • 8/30/2017 6:59:44 PM
NO RATINGS

@Ariella: Which goes to my main point when I get in conversations like this: measuring fitness for a job by degrees and formal education is an obsolete, losing proposition in this day and age for most jobs/organizations -- especially what with all of the online-learning resources available to the motivated and the quick.

Re: Subjectivity and Algorithm's Can't Play Nice.....
  • 8/29/2017 2:51:35 PM
NO RATINGS

@Joe yet I see it all the time. Of course, they would claim they need that data as way of checking if the education you claim is correct. One time, when Pearson was on its outsource personnel info kick, there was a real problem because the online form had a drop down menu that omitted many universities, including the one I attended for grad school. But anyway, once they ask for years, they could easily come up with an estmate. In my case, they'd be off and think I'm older than I actually am.

Re: Subjectivity and Algorithm's Can't Play Nice.....
  • 8/29/2017 12:57:29 PM
NO RATINGS

@Ariella: Whether or not they would win or not (and I certainly don't propose to give legal advice or create/affirm/implicate/acknowledge an attorney-client relationship here in an Internet comment), this sounds like inviting a lawsuit... Almost certainly not best practice from a compliance perspective.

Re: Subjectivity and Algorithm's Can't Play Nice.....
  • 8/26/2017 10:37:14 PM
NO RATINGS

No doubt we are all hoping to accumulate extra dollars on our paycheck each year as we get older. Come to think of it, i think we all should hope to be able to move on to the freelance / contract economy by the time we are old enough for people not to want us full time. No?

Re: Subjectivity and Algorithm's Can't Play Nice.....
  • 8/26/2017 10:13:31 PM
NO RATINGS

<? At that point, you have perhaps reached max saturation of knowledge and experience, so the only thing you're accumulating is white hairs and wrinkles.>

LOL @broadway Indeed. The only thing they may say is that in some sectors, the expectation is built in to get an increase in salary every year. In such cases, a few more years expereince could cost a few thousand dollars more in salary. But that is not all that common outside unionized type positions.

Re: Subjectivity and Algorithm's Can't Play Nice.....
  • 8/26/2017 10:11:16 PM
NO RATINGS

<That's the other thing. Leave the year of your degree(s) off of your resume/LinkedIn. ;)>

@Joe I do that. However, the online applications I've seen usually demand specifics not just for the degree date but the beginning of enrollment, as well. 

Re: Subjectivity and Algorithm's Can't Play Nice.....
  • 8/26/2017 9:53:06 AM
NO RATINGS

@Broadway: Well, thanks for making us all feel better about getting older! :p

Re: Subjectivity and Algorithm's Can't Play Nice.....
  • 8/25/2017 10:07:37 PM
NO RATINGS

There's no other way to describe their approach toward a maximum experience level. Really, what's the difference between someone with 20 years experience and 25? Or 27 years? At that point, you have perhaps reached max saturation of knowledge and experience, so the only thing you're accumulating is white hairs and wrinkles.

Re: Subjectivity and Algorithm's Can't Play Nice.....
  • 8/25/2017 7:25:33 PM
NO RATINGS

@Predictable: There could be some hidden age discrimination going on there. Who knows? On the one hand, "overqualification" is a legitimate reason (legally speaking) to rule someone out -- but there is arguably a thin line between overqualification and age discrimination. Is there really that big of a difference between 22 years experience and 25 years experience? One could make some compelling arguments either way, I suspect.

(DISCLAIMER: Not legal advice. Neither this nor other posts/comments here constitute legal advice or the creation, implication or confirmation of an attorney-client relationship. For actual legal advice, personally consult with an attorney licensed to practice in your jurisdiction.)

Re: Subjectivity and Algorithm's Can't Play Nice.....
  • 8/25/2017 7:21:41 PM
NO RATINGS

> Then the question would arise: what did you do all those years between earning your degree and beginning at this job?

That's the other thing. Leave the year of your degree(s) off of your resume/LinkedIn. ;)

Page 1 / 4   >   >>
INFORMATION RESOURCES
ANALYTICS IN ACTION
CARTERTOONS
VIEW ALL +
QUICK POLL
VIEW ALL +